Volume 11, Issue 1 - Proceedings of the 2025 Joint Annual Conference of the Austrian (ÖGBMT), German (VDE DGBMT) and Swiss (SSBE) Societies for Biomedical Engineering (9–11 September 2025, Muttenz/Basel, Switzerland)

Introduction: Energy modulators (EM) have gained increasing attention due to their relevance in FLASH proton therapy. Current EM designs are fragile, necessitating thin ridges and limited sizes. Production requires advanced and costly 3D printers, as basic filament printers are inadequate, leading to expensive outsourcing. We propose an innovative EM design to address these limitations. Methods: The novel design consists of boxes stacked one on top of the other. Each box is created with two distinct infillratios: one for the inner cube and one for the larger cube surrounding it. The size of the inner and outer cubes is optimized along with the infill ratio. This structure is repeated periodically for box-like targets or varied to achieve the desired dose distribution.We design, slice, and print the EM using a Bambu Lab-X1C filament printer and polylactic acid filament. The simulation was performed using a GPU-based Monte Carlo simulation (FRED), where the EM was modeled as a combination of cubes with variable heights, with the infill ratio mapped to the material height. The simulation design was validated through depth-dose curve measurements using PSI Gantry 2. Results: EM for box-like targets of 35mm depth were designed and manufactured by varying the infill ratio between 30% and 100%. We chose cubic infill for isotropic periodicity, ensuring that impinging particles encounter uniform material distribution. Manufacturing constraints require 100% (concentric) infill in areas smaller than 1.5mm2. The simulations were validated by depth dose curve measurements with shape and distal range agreement to within detector uncertainties (<1mm). Conclusion: We realized a 3D-printed EM that is easier to produce and more resistant to physical damage than current alternatives. This design demonstrates the potential for a costeffective, easily manufacturable, and scalable EM.

Open-porous structures produced by Additive Manufacturing (AM) offer an excellent basis for patient-specific implants, scaffolds and medical devices. The interface of 3D structures produced by this technology can be improved by the thin-film technology of Atomic Layer Deposition (ALD), a highly modified form of Chemical Vapour Deposition. ALD allows to apply highly conformal coatings on complex geometries and coating of internal surfaces and channels. ALD is a versatile method, as it is compatible with AM. In this work, the conformity of the ALD coating on open-porous structures was investigated. Lattice type cubes of 10x10x10 mm3 outer size and with square channels in all three orthogonal directions with widths of 2’000 μm and 750 μm and two, four or six open sides were produced using masked stereolithography. These additive manufactured open-porous structures were coated with a 75 nm thick titanium oxide (TiO2) layer by ALD. The TiO2 coating thickness was determined on different positions within channels using Scanning Electron Microscopy (SEM). The measurements revealed excellent coating conformity over the internal surfaces (82.1 ± 2.6 nm and 87.9 ± 3.8 nm in 2’000 μm and 750 μm channels, respectively) with a Uniformity Index of up to 7.8%: The results highlight the potential of combining 3D printing and ALD to overcome existing technical limitations and enhance the functionality, biocompatibility, and durability of advanced materials for medical applications.

Tubular Structures (TS) play a central role in the human body due to their wide range of applications. Many native structures consist of TS with different properties. These structures fulfill a variety of biological and mechanical functions. In some cases these structures need to be replaced. SMES (melt electrospinning) offers a process to manufacture artificial TS. The employed 3D printer (X350pro, German RepRap), an SMES process, is modified by connecting a negative highvoltage source to a collector on the print bed while the nozzle is grounded. For the manufacture of TS the collector is a rotating metal tube. A Scanning electron microscope (SEM) (TM-1000, Hitachi) is used to obtain information on the general fiber topology and, in particular, the fiber bonding. The fiber diameters are determined with a fiber analysis program (MAVIfiber2d, Fraunhofer Institute) using SEM images of sputtered (Coater SCD 040, Balzers Union Limited) samples. The topography change of TS by varying the extrusion multiplier and the rotational speed is examined macroscopically. In order to evaluate the process stability of the build-up, a print series is examined. For the optimized standard print, reliably closed TS can be produced with an average fiber diameter in the range of 30.43 to 35.28 μm. The fiber topology can be modified by changing parameters like speed and extrusion multiplier. The developed manufacturing process for tubular nonwoven structures thus offers a versatile manufacturing process that can be investigated in more detail for specific TS in medical applications.

Microphysiological systems (MPS) have emerged as a valuable tool in preclinical drug testing. These systems are capable of simulating in vivo conditions and physiological functions within an ex vivo tissue culture setting, resulting in improved accuracy of the obtained data. However, the labscale manufacturing processes of MPS are associated with high manufacturing time and costs, and are thus unable to sustain high-throughput studies [1]. Thus, we have to rethink the design of MPS considering high-volume production technologies from the beginning on. In this study, we examined which design features can help to translate the production process of MPS from a small scale micro-milling process to high-volume injection moulding. Moreover we compared the manufacturing costs of both manufacturing processes to figure out the right time to switch to injection moulding. To compare both manufacturing processes, we designed a tissue culture MPS by using common design features that fit to both manufacturing processes and compared design features that have to be changed, when transitioning from micro-milling to injection moulding. The break-even point was investigated using public pricing tools suggesting, that injection moulding has considerable potential in enhancing the scalability of MPS production starting from a production volume of 200 units. This could help further projects reducing manufacturing costs and time and foster the early switch to injection moulding, thus yielding in an enhanced accessibility of new MPS concepts.

Progressive Addition Lens (PAL) exhibits complex freeform surfaces. More and more user specific optimization is designed into modern PAL, increasing the complexity of its characterization. Also, modern additive manufacturing methods implement not only ways to create individual freeform shape, but to vary locally refractive index, too. Improvements demand complementary methods to topography measurement and traditional powermapping to characterize the optical performance, especially in a usercentric way taking into account individual direction of sight. We present an overview of different implementations of experimental raytracing (ERT) for measurement of relevant functional parameters of PAL and contact lenses (CL), locally resolved, and taking into account different illumination scenarios, simulating real life use cases.

The paper examines the process parameters of 3D printing for the manufacturing of implant structures made of calcium phosphate cement, taking into account different overhang angles. The aim of the investigation is to ensure a dimensionally stable manufacturing of overhangs that meets the requirements of individual implants. To this end, samples with and without an outer contour were printed at different angles (20° to 48°) and with different numbers of layers (16, 24, 32, 40) and analyzed metrologically. The methodology included a systematic variation of relevant parameters to investigate the influence of the overhang angle and the number of layers on the shape deviations. The results showed that samples without an outer contour still exhibited sufficient dimensional stability up to an overhang angle of 38°, while significant dimensional distortions were observed in samples with an outer contour from an angle of 30°. In both cases, a layer count of 24 proved to be optimal for high overhang quality.

Plastics are crucial in modern medical technology due to their adaptability and cost-effectiveness, but they also produce substantial waste. This study investigates the effects of recycling and γ-irradiation on stabilized polypropylene copolymer (PP) containing the secondary antioxidant Irgafos 168 (I168), the acid scavenger calcium stearate (CaSt), and the radiation stabilizer Tinuvin 622 (T622). Using a Design of Experiment (DoE) approach, it evaluates how processing, irradiation, and radiation stabilization interact, focusing on I168 degradation and rheological changes. When comparing the first and the third processing and γ-irradiation cycle, the I168 content and the formation of the radiolysis product 1,3- di-tert-butylbenzene (DTBB) were measured by highperformance liquid chromatography (HPLC) and gas chromatography-mass spectrometry (GC/MS). γ-Irradiation significantly reduced I168 content and lowered the viscosity of the PP. Although T622 mitigated effects, it didn't fully prevent I168 degradation. Processing influences were less significant compared to ionizing radiation. The study highlights the necessity for radiation stabilization during sterilization and suggests adjusting I168 post-irradiation for recycling. Continuous monitoring of degradation products is recommended to maintain product safety thresholds.

The intersection of 3D printing technology and materials science has catalysed the development of bioactive materials with tailored properties, crucial for expanding applications from temporary to permanent components in fields such as dentistry and tissue engineering. This study explores the formulation and characterization of a 3Dprintable resin comprised of methacrylate-based compounds filled with hydroxyapatite (HAp) and chlorhexidine (CHX). By integrating HAp and CHX, the resin aims to synergize antimicrobial activity with osteoconductivity, potentially enhancing the material's utility in biomedical applications by promoting bone cell attraction and minimizing inflammation during healing. Dimensional accuracy tests were conducted on rectangular and circular structures, revealing acceptable accuracy. HAp content was effectively quantified using a complexometric titration post-incineration. CHX extraction, despite attempts with methanol and acetic acid in both ultrasonic and Soxhlet methods, recovery rates did not exceed 55.3%, indicating significant retention within the resin possibly due to intermolecular interactions with the macromolecules. This research establishes a robust methodology for the incorporation of bioactive agents in 3Dprinted materials, offering insights into optimizing resin formulations. By addressing these challenges, the study contributes to advancing 3D-printed resin technologies with potential long-term applications in the biomedical sector.

Periodontitis is a prevalent chronic inflammatory disease that requires effective long-term treatment strategies. Current antimicrobial formulations, such as Chlorhexidine (CHX) gels and antibiotic-based inserts, often suffer from short retention times and uncontrolled drug release, limiting their therapeutic efficacy. To address these challenges, we developed a biodegradable, in situ cross-linked CHX depot incorporating a bifunctional reactive cross-linker, triethylene glycol bis(dilactic acid) bis(hexamethylene diisocyanate) (TELANCO). The synthesized TELA-NCO was mixed with commercially available CHX gel to form an injectable paste that polymerizes within 7 hours into a stable depot. HPLC release studies demonstrated sustained CHX release for over 35 days, significantly reducing initial burst effects. In vitro cytocompatibility tests revealed expected concentrationdependent CHX cytotoxicity, but good biocompatibility of the cross-linker. This injectable, degradable depot system offers a promising alternative for localized antimicrobial therapy in periodontology and endodontics, eliminating the need for secondary removal.

Using Size exclusion chromatography (SEC), the molar masses Mn and Mw in a binary polymer mixture with polyethylene glycol are to be determined in order to find a mathematical relationship at different PEG contents. Based on this, a quantitative PEG determination in samples with unknown PEG content should then be enabled. It is shown that this mathematical correlation is only found as far as polydispersity is taken into account. The investigations were carried out for two biostable polyurethane-based polymers as well as for three degradable polymers in order to demonstrate the general validity of the results.

Conventional transvenous rhythm implants are still the treatment of choice for patients with cardiac arrhythmias. Complications are frequently associated with the polyurethane insulation of intracardially implanted pacemaker leads. For identification of insulation materials with improved properties, a suitable in vitro test setup is necessary. The present study focusses on the development of an in vitro oxidation model for analyzing the biostability of insulation materials under accelerated biochemical deggradation mechanisms. A degradation of cylindrical Pellethane 80AE test specimens was conducted using three oxidizing test solutions (OS) of hydrogen peroxide and cobalt(II) chloride hexahydrate at 22°C, 37°C and 40°C. Surface morphological and structural changes of the insulation materials were analyzed in comparison to an untreated reference. Scanning electron microscopy showed degradation defects in the insulation surface of Pellethane 80AE, increasing with increasing concentrations of hydrogen peroxide and cobalt(II) chloride hexahydrate of the OS. Likewise confocal laser scanning microscopy revealed an increase of surface roughness with higher concentration and temperature of the OS. Evaluation of the chemical surface structure showed oxidatively induced changes in the soft segments representing known effects of polyurethane degradation. Thus, for a simulation of biochemically induced changes in the biostability of lead insulation materials such as Pellethane 80AE, an oxidizing test solution with a higher concentration of hydrogen peroxide at a test temperature of least 37°C is suitable to provoke significantly accelerated chemical degradation.

Postoperative scarring following trabeculectomy, a glaucoma surgery, often leads to failure, necessitating biodegradable drug delivery systems for sustained antifibrotic release. This study explores γ-cyclodextrin/hyaluronic acid (γ- CD/HA) copolymers as potential drug depots, evaluating their degradation kinetics under physiological conditions. Copolymers with 0% and 75% γ-CD were synthesized via base-catalyzed crosslinking with 1,4-butanediol diglycidyl ether (BDDE). Degradation was assessed by incubating samples in Sørensen buffer (pH 7.4) with lysozyme, monitoring mass loss over 14 days. Both systems degraded, yet the 75% CD copolymer exhibited a significantly slower, linear degradation profile compared to the rapidly degrading 0% CD system. Morphological analysis revealed increased swelling in the 75% CD samples, suggesting cleavage of intermolecular crosslinks. Gravimetric analysis indicated a mass loss of approximately 50% for the 75% CD system by day 14. The tunable degradation kinetics of these γ-CD/HA copolymers suggest promise for controlled drug release applications in glaucoma, warranting further investigation into drug loading and release profiles.

An atraumatic implantation of the Cochlear Implant (CI) to prevent, inter alia, loss of spiral ganglion neurons (SGN) and residual hearing, might be achieved by a flexible, smooth, and hydrophilic coating of the electrode. Such a coating needs to be long-term stable to avoid swelling or shrinking, which cause micro-movements of the electrode. In parallel, it has to be harmless to hair cells (HC) and SGN. Due to its inert and biocompatible character, barium cross-linked ultra-high viscosity alginate is a promising candidate for this propose and was tested for stability and biocompatibility under simulated inner ear conditions. CI dummies were coated with alginate by using a cone-shaped semipermeable membrane. Alginate sol was injected in these membrane cones, dummies inserted, and all covered with barium cross-linking solution for ionic gelation. Coated dummies were incubated in artificial perilymph and observed for one year with weekly medium change and photo documentation for stability assessment. Additionally, beads of alginate hydrogel were co-cultivated with murine inner ear tissue for one week to test biocompatibility with SGN and HC. Alginate hydrogel coating stayed attached over the one-year observation period. Small variations in the diameter of coated dummies were detectable, likely due to rotation of the floating dummies, but no clear shrinking or swelling was seen. Neither number of SGN nor number of HC was significantly reduced by co-culture with alginate beads. In simulated inner ear conditions, alginate hydrogel was stable as coating for one year without greater changes of shape and had no neuro- or ototoxic effect. These promising results support testing of alginate as lubricant-coating in vivo as a further step towards translation to clinical use for CI improvement.

Stents play a vital role in the treatment of cardiovascular diseases. In this context, endothelialization is of crucial importance for the final implant performance and the key to an effective treatment. To enhance endothelial cell (EC) attachment and surface coverage, laser-induced periodic surface structures (LIPSS) have been identified as an effective surface modification technique. This study examines the effects of groove-like and spike-like LIPSS on EC behavior using femtosecond laser-machined stainless steel substrates. The fabricated structures were evaluated with respect to EC orientation and morphology through in vitro experiments. The findings indicate that groove-like LIPSS notably impact EC orientation, demonstrating their ability to enhance endothelialization. Additionally, a stainless steel stent featuring groove-like LIPSS was produced, illustrating the feasibility of incorporating LIPSS into industrial fabrication processes for cardiovascular implants. This study underscores the potential of LIPSS to enhance stent biocompatibility and long-term performance.

To study the effects of metal ions on cells and tissues caused by wear and corrosion of implants in vitro, a system is required that releases metal ions in a controlled and sustained manner. As small concentrations can have an impact on biological systems, a correspondingly slow release over a defined period of time is required. A method for the manufacturing of polymeric poly(L-lactic acid) (PLLA) films with incorporated cobalt microparticles by dip coating is reported. Two analytical methods are established to demonstrate the detection and quantification of cobalt particles. Thermogravimetric analysis allows the quantification of cobalt particles. Scanning electron microscope images can be used to determine the distribution and number of particles of different size ranges.

Magnetic hydrogels are emerging as promising biomaterials for tissue engineering due to their ability to provide structural support and controlled drug release. In this study, gelatin-hyaluronic acid hydrogels (70:30) were synthesized and functionalized with Divi-Divi (Libidibia coriaria) extract (1% w/v) and magnetic nanoparticles (Fe3O4@SiO2, 1% w/v). The extract was obtained through an aqueous extraction method and characterized for its bioactive properties. Rheological analysis showed that the incorporation of magnetic nanoparticles increased hydrogel stiffness, whereas Divi-divi extract did not significantly affect the mechanical properties. Controlled release experiments were conducted using a 30 mT magnetic field at 100 kHz for 30 minutes in PBS, resulting in a release of approximately 62 μg/ml of the extract, as detected by UV spectrophotometry. These results demonstrate the potential of these hydrogels as magnetically responsive drug delivery systems for biomedical applications. Future studies will optimize the release kinetics and evaluate their biocompatibility in vitro.

Nanofiber nonwovens are of increasing significance in the field of biomedical applications. Knowledge of material characteristics is a fundamental requirement for the development of safe medical devices. A variety of suitable methods are required for process monitoring and to analyze material parameters. We consider and reflect specific features of selected methods like microscopy, scattering techniques or porosimetry. Potential methodological demand is discussed. A brief literature based overview is provided regarding to characterization methods for in situ or post process routines. Upcoming demands are derived from existing gaps in the spectrum of methods. Possible potential concepts for new optical techniques that can be designed and implemented in prospect technologies are introduced and discussed. The authors are actively working on method development regarding to material characterization and process monitoring.

The clinical picture of crytorchidism describes a deviation in the position of the testicles outside the scrotum and is one of the most common anomalies in male newborns. The production of customized anti-adhesion membranes as artificial biomaterial pouches seems to be a key element in the temporary isolation of the testicles during two-stage orchiopexy. In this work, we produced electrospun PLGA scaffolds with and without lecithin to compare this highly porous fibrous membrane with commerically available ePTFE and biological matrices. First, the surface morphology of all specimens was assessed by SEM. In addition, their wettability, mechanical behaviour, and biological response were considered in an initial screening. Cell tests over 3 days using a human fibroblast cell line showed the potential functionality of lecithin within electrospun fiber nonwovens and promising results for the tissue matrix, which still need to be confirmed in future studies. The next step will focus on further reducing adhesion of specific cells for testicular pouch application.

A novel phenylalanine/ valine ethyl ester cosubstituted polyorganophosphazene, incorporating triethyl citrate as plasticizer, is being investigated for its degradability. Films were solvent casted, and allowed to degrade for 4 weeks at an elevated temperature of 50 °C in water and two HEPES buffered solutions with pH values 6.8 and 7.4, respectively. The degradation of the samples in pure water occurred at a significantly accelerated rate compared to the samples incubated in HEPES buffer. Furthermore, based on their influence on degradation, the pH values of the media used were assessed.

Nosocomial pathogens spread in hospitals and are a permanent issue to patient safety [1]. Disinfecting medicinal equipment is therefore most important. Usually, chemical disinfection is applied but time consuming. Another known disinfection technique is irradiation with wavelength below 280 nm (UVC). UVC efficiently inactivates pathogens but usually cannot be applied in the presence of humans. The recently emerging Far-UVC with wavelengths between 200 and 230 nm is assumed to pose no danger to humans and is considered to be applied more frequently in the future. Therefore, this study investigates the potential radiation resistance development of Escherichia coli K12 caused by repeated exposure to 222 nm Far-UVC irradiation and the persistence of these adaptations. A cyclic irradiation experiment was conducted over ten cycles, followed by the assessment of resistance persistence over four subsequent cycles. The sensitivity of bacteria to irradiation was analyzed for exposure durations of 5, 10 and 15 seconds at an irradiation intensity of 1 mW/cm2. The number of the bacteria after the incubation between the cycles after a fixed incubation time and morphological changes in bacteria colonies on Luria-Bertani (LB) plates were observed, indicating potential adaptation mechanisms. The results indicate that cyclic irradiation with this amount of irradiation intensity and the duration of several seconds does not induce significant and persistent Far-UVC radiation resistance but suggests transient adaptive responses which are not permanent. Adaptive reactions must be investigated more closely to avoid creating additional radiation resistance to existing drug resistances

Polymeric per- and polyfluoroalkyl substances (PFAS), such as polytetrafluoroethylene (PTFE) and its expanded form (ePTFE), are widely used due to their exceptional chemical resistance and biocompatibility, particularly in medical devices. However, environmental and health concerns linked to PFAS persistence and contamination have prompted the European Union to propose restrictions on their non-essential uses. This has raised challenges for industries dependent on PFAS-containing materials, especially in biomedical applications. In response, this study explores polyurethane (PU) as a potential alternative to ePTFE. Three PU variants were evaluated for their mechanical and biological properties, aiming to identify suitable substitutes that maintain the performance required for critical medical applications.

This paper introduces a framework for estimation of lower-leg biomechanical parameters (inertia, damping, stiffness) using a passive exoskeleton with embedded IMUs and position encoders. By combining Kalman filtering with sequential least squares optimization, we achieve subjectspecific identification during natural movement without motion capture systems.

Introduction: Generative Design (GD) is an intelligent process that uses algorithms to automatically create optimised, functional and material-efficient structures based on defined parameters and loads. This makes it a valuable tool for the customisation of orthopaedic insoles to suit individual load patterns. Conventional methods are frequently predicated on pressure data. However, force-based modelling has the potential to reproduce dynamic gait phases with greater accuracy. Additive manufacturing facilitates the production of such complex, customised designs. Methods: In this preliminary study, a comparison is made between different load definitions (pressure vs. force) and manufacturing constraints (additive vs. unrestricted) within GD. Pedographic pressure data is utilised to ascertain individual load distributions. The maximum value recorded during walking was utilised for the purpose of analysing the pressure load. In order to analyse the effect of force, a separate load case was defined for each gait phase, with the objective of mapping the application of force along the gait line as precisely as possible. Results: Within GD, forces can be positioned specifically for each gait phase, allowing structures to be generated with a precise fit. Conversely, pressures act over the entire surface and overlap when analysing multiple gait phases. The findings demonstrate that force-based modelling offers a more precise depiction of dynamic loads, consequently resulting in optimised insoles. Additive manufacturing constraints prove beneficial by removing the need for internal support structures, which are often hard to remove. However, thickened structures in the upper layer of the insole can compromise flexibility. Conclusion: The employment of GD facilitates the development of orthopaedic insoles that accurately reflect the patient’s individual load distribution. Future research should investigate the impact of different design choices on flexibility and biomechanical performance. In addition, the identification and evaluation of suitable materials, is crucial for determining their effectiveness in the manufacture of foot orthoses.

The permeability properties of covered stents are crucial for the therapy success treating peripheral artery disease, peripheral artery occlusive disease as well as dissections and aneurysms. For investigation of water permeability according to ISO 7198:2016 a customized test setup was developed. It enables for standardized testing regarding the integral water permeability as well as the water entry pressure and provides additional quantitative and qualitative results beyond the requirements: The first measureable flow, the maximum applicable pressure as well as photo documentation during the test were implemented. Four commercially available covered stents indicated for atherosclerotic lesions of the iliac arteries were investigated within the current study showing no measureable permeability at 120 mmHg. However, different water entry pressures and leaking patterns at higher pressures could be observed. Especially, photo documentation during the test as well as the first measureable flow enhance the informative value of standardized permeability testing.

The development of stent technologies faces new challenges with advent of polymeric materials, offering versatility in mechanical, physicochemical, and biological properties. Biodegradable polymers, such as poly-L-lactide (PLLA), are particularly promising. Numerical simulations, particularly finite element analysis (FEA), play a crucial role in optimizing stent designs. However, validating advanced material models for FEA is essential due to the lack of consensus on data collection and validation methods. This study introduces a validation framework using planar stent segment expansion to assess the mechanical behavior of polymeric materials and designs efficiently without extensive stent production. PLLA specimens were manufactured via injection molding and tested at 37 °C. Two material models, a linear elastic-plastic (LEP) model and an advanced viscoplastic (AVP) model, were compared. In summary, AVP showed improved conformity with experimental data. Our results demonstrate the importance of advanced material models for accurate FEA simulations and bridges the gap between numerical simulations and experimental validation in polymeric stent applications.

The ISO 7198:2016 standard requests permeability testing for medical cover materials but does not consider for physiological viscosity. This study examines the impact of fluid viscosity on the permeability of polymer membranes under static pressure conditions. A specialized test bench was used to evaluate the permeability of two membrane materials with water and a water-glycerol mixture. The appropriate water-glycerol ratio was determined by rheometric analysis to match physiological viscosity. Permeability was assessed at four static pressure levels using a flow sensor, following ISO 7198:2016. Permeability increased with pressure for both test fluids. However, permeability with the water-glycerol mixture was significantly lower than with water. The application of water alone for permeability testing may not represent the materials behavior under physiological conditions. Including permeability measurements with fluids of physiological viscosity could improve the reliability of standardized testing, ensuring more accurate material characterization for medical applications. On the other hand, the use of pure water represents a worst-case scenario if low permeability of the membrane is required.

Involuntary childlessness is a deeply sensitive issue. In one-third of cases, female infertility is attributed to occlusions in the Fallopian tubes. Reproductive treatment methods are often significantly physical, psychological, and financial challenging. Our previously introduced microstent technology offers an innovative therapeutic option for restoring the lumen of the Fallopian tube without the need for surgical interventions. A crucial requirement for the success of such a device is its ability to endure the continuous stress generated by the regular contractions of the tubal musculature. This contribution aims to explore the durability of the device using a novel experimental setup that cyclically compresses the microstent prototype between two parallel plates. The number of load cycles varied between 0 load cycles (corresponding to 0 days) and approximately 2,000,000 load cycles (corresponding to 365 days). After completing the load cycles, an analysis of the height/width ratio of the microstent prototypes was conducted and the samples were analyzed by scanning electron microscopy with different zoom levels. As the number of load cycles increases, the ratio of height to width decreases. First signs of degradation were documented after 500,100 load cycles corresponding to a time period of 90 days in vivo. Despite the advanced degradation, no fractures or cracks were documented.

Acute type A aortic dissection (ATAAD) remains a life-threatening condition with high morbidity and mortality rates despite advancements in surgical and perioperative management. The Ascyrus Medical Dissection Stent (AMDS) has been introduced as an adjunct to the hemiarch procedure to reduce complications such as true lumen collapse and false lumen patency. Knowledge of radial force behaviour, is essential when selecting endovascular stents, as it ensures vessel support and maintains true lumen patency. This study compares the mechanical behavior of the AMDS to conventional aortic stent grafts, focusing on differences in radial force characteristics. Using the Jotec E®-VITA Thoracic 3G stent graft as a reference, radial force measurements were taken using a segmented head testing machine, and axial elongation was measured at corresponding diameters. While the absolute chronic outward force (COF) increases similarly for all devices during compression, the AMDS 55-40 exhibits the highest COF, followed by the Jotec device and the AMDS 40. When normalizing force to device surface area (COF/A), the Jotec device shows a steeper increase, indicating that its impact on the vessel wall is more dependent on vessel diameter. In contrast, the AMDS devices maintain a more consistent COF across different diameters, suggesting a broader application range in terms of aortic diameters.

Advanced technologies help in the analysis of the surface electromyogram (sEMG) signals to recognize the pattern of the gait and provides the mechanism to control the prosthetics. EMG signals used for the diagnosis of neuromuscular diseases, human-machine interaction as they reflect human intensions. In this study an attempt has been made with novel approach for the classification of walking activity using a hybrid CNN-LSTM model based on the sEMG and goniometer (GM) signals from healthy and knee abnormal subjects. This study utilizes the dataset from UC Irvine Machine Learning Repository (UCI) on lower limb sEMG signals of 11 normal and 11 abnormal subjects with knee abnormality while performing the walking activity. A hybrid CNN-LSTM effectively captures the temporal and spatial features from sEMG signals and provides the average accuracy and average precision of 82.10%, 72.69%. This performance suggests that the proposed model successfully classifies the sEMG signals during walking and thereby leading to more precise control of the prosthetics using the hybrid model.

This study evaluates the biomechanical and optical consequences of three corneal refractive surgeries-PRK, LASIK, and KLEx-using patient-specific finite element simulations. Personalized corneal geometries were derived from preoperative tomography of 30 patients and subjected to virtual interventions modeled in a nonlinear anisotropic framework. Material parameters were calibrated from experimental human corneal data. Stress distributions and refractive outcomes were computed under physiological intraocular pressure. Despite identical tissue removal volumes, PRK preserved a more uniform stress distribution and delivered the least optical undercorrection. LASIK induced significant posterior stromal stress and KLEx, although mechanically intermediate, exhibited greater undercorrection due to cap-induced tension. The results support the integration of biomechanics-informed modeling in refractive surgery planning.

Proximal Fallopian tube occlusion represents a significant cause of female infertility, limiting natural conception. To address this challenge, we proposed the FemStent previously - a polymer-based, bioresorbable, and self-expanding microstent. To accelerate the epithelialization of the inner stent surface and thus enable safe and uninterrupted passage of gametes, the concept of microstructuring was developed. Corresponding patent applications have been filed (EP 23200861, PCT/EP2024/077397). The current study analyzed the technical feasibility of microstructuring the inner stent surface. For this purpose, stainless steel mandrels were microstructured using femtosecond laser ablation and then used to manufacture tubular semifinished products made of poly-(L-lactide). Microstent prototypes were manufactured by using femtosecond laser cutting of polymer semifinished products with and without a microstructured inner surface. The microstent prototypes were subjected to comprehensive morphological analyses. An in vitro model system for particle tracking was developed to demonstrate the permeability of the fallopian tube stent for gametes. The results demonstrate the technical feasibility of a fallopian tube microstent with a clearly defined microstructured inner surface and provide evidence for improved gamete transport. Future in vitro studies in more elaborate models as well as future in vivo studies will provide further insights into optimizing the microstent's structure and function. This research lays the groundwork for an innovative gynecological intervention, potentially improving fertility outcomes in patients with proximal tubal occlusions in the future.

Transcatheter Aortic Valve Replacement (TAVR) is a widely used treatment for severe aortic stenosis, particularly in patients with a high surgical risk. Although initially designed for tricuspid aortic valves (TAV), its use in bicuspid aortic valves (BAV) presents unique challenges due to anatomical variations and extensive calcifications, which can hinder optimal stent expansion. In this study, a prototype of an elliptical TAVR stent was produced. Radial force measurements were carried out to calibrate the simulation model. In addition, the deployment behaviour of this elliptical TAVR stent was investigated in a patientspecific bicuspid aortic valve geometry with severe calcification. Finite Element (FE) simulations were conducted to analyze the crimping and expansion process of an elliptical TAVR stent. Expansion behavior and radial force behavior were compared with a conventional circular stent design. The simulation results demonstrate that both stents exhibit elliptical deformation due to calcification-induced constraints. However, the elliptical stent experiences less deformation in the narrowest regions, potentially reducing leaflet stress and improving valve function. The elliptical design may enhance the longterm durability of the artificial valve. The findings highlight the potential benefits of elliptical TAVR stents for bicuspid anatomies. Further investigations, including large-scale in silico studies are necessary to optimize stent design and improve clinical outcomes for bicuspid valve patients.

Minimally invasive transcatheter aortic valve replacement (TAVR) has become the preferred procedure for patients with aortic valve stenosis or insufficiency who are at high risk for conventional open-heart surgery. The favorable clinical outcomes observed in high-risk patients have led to an expansion of the eligible cohort, now including intermediateand low-risk patients. A critical aspect of advancing TAVR procedures lies in preoperative simulation, which in future can integrate patient-specific in-silico deployment simulations and post-deployment fluid mechanics assessments. In light of the aforementioned context, this study investigates the volumetric distribution of calcifications in the aortic cusps of a TAVR patient population and explores whether naturally occurring clusters should be considered in in-silico simulations. We analyze clustering results based on different methods for feature extraction, using circular measured volumetric calcification distributions. The clustering method applied is hierarchical clustering. Our findings identify distinct calcification clusters across different feature extraction methods. The Framework can be incorporated into in-silico trials and clinical studies assessing the impact of calcification patterns on clinical outcomes and TAVR device optimization.

While MRI is the modality of choice for soft tissue imaging, its ability to visualize bone structures remains limited. In contrast, CT provides superior bone imaging but involves ionizing radiation, making it suboptimal for longitudinal studies in pediatric populations. This study investigates the feasibility of using Ultrashort Echo Time (UTE) MRI sequences to visualize vertebral bone anatomy and support reliable morphometric assessments over time. Three adolescent volunteers (mean age: 15 ± 1 years) underwent MRI scans including two UTE sequences of the thoracic and lumbar spine, acquired at two time points within a single imaging session. Registration between time points was evaluated using the Dice similarity coefficient and surface distance analysis, considering both 95th and 99th percentile thresholds were calculated. The segmentation achieved high repeatability with an average Dice coefficient of 0.92 ± 0.02. Mean surface distances between time points were 0.44 ± 0.08 mm at the 95th percentile and 0.50 ± 0.14 mm at the 99th percentile. Region-specific results showed slightly better agreement in the lumbar region (Dice: 0.94 ± 0.02) compared to the thoracic spine (Dice: 0.92 ± 0.02). UTE MRI provides enhanced bone contrast and enables accurate, repeatable vertebral segmentation. The observed surface distances fall below the expected yearly vertebral growth in healthy children (0.8-1.2 mm/year), indicating that UTE-based segmentation is sufficiently precise for longitudinal assessment of vertebral development without the risks associated with radiation exposure.

Reflection‐based 3D scanning techniques are widely used in medical applications today. In these techniques, encoded light patterns are projected onto the skin, reflected, and captured by 2D cameras. Conventional RGB white‐light projectors emit wavelengths that are suboptimal for darker skin types. This results in weakly reflected patterns, prolonged exposures, and increased artefacts in the 3D image. The aim of this project is to optimise an RGB projector for reflectionbased 3D scanning of human skin. The methods included evaluating the intensity of reflected patterns and the LED power consumption. A novel concept employing three red LEDs with slightly shifted wavelength peaks was devised and simulated. It exploits the ideal spectrum, while keeping costs and power consumption low.

Radiopacity is an important prerequisite for X-ray visibility of implants. Existing standards refer to the use of Xray systems for medical imaging. Micro Computed Tomography (μCT) technology is available with high resolution imaging. Experimental studies should verify the comparability of radiopacity results with respect to filtering, image resolution and composition of reference samples. The study yielded that radiopacity can be measured with differing equipment. Advantageous for μCT are high resolution, flat field correction and avoidance of radiation protection measures when using full protection devices. However, the results cannot match the requirements of related standards, because sample distance and simulation of body scattering are limited. Both might be important when using the data for assessment of X-ray visibility in clinical settings.

Clinical computed tomography (CT) plays a critical role in patient-specific implant planning; however, its standard resolution is often insufficient for precise biomechanical simulations.We suggest a multi-planar reconstruction pipeline that fuses three image-plane reconstructions to increase the effective resolution. Clinical CT datasets of human femoral heads are processed by converting DICOM images into the Visualization Toolkit (VTK) format, followed by voxel-wise simple mean intensity integration to enhance the image resolution. Quantitative evaluations using intensity profiles and statistical analysis confirmed that the proposed method not only preserves global consistency but also reveals essential local variations in bone structure. The enhanced resolution facilitates more accurate geometric extraction and material mapping for subsequent finite element analyses.

Magnetic Drug Targeting (MDT) is a promising technique for local chemotherapy, employing magnetic nanoparticles (MNPs) and an external electromagnet to concentrate chemotherapeutic agents at tumor sites. Achieving complete tumor perfusion with the drug requires precise control over MNP distribution, which can be optimized by repositioning the electromagnet. However, this process needs a therapy monitoring system capable of 3D mapping of MNP distribution within the tumor. While ultrasound-based imaging techniques, particularly magnetomotive ultrasound (MMUS), have been shown to effectively track MNP accumulation in 2D, a full 3D characterization remains challenging. In this study, we employ a matrix transducer combined with a sliceby- slice 2D global ultrasound elastography (GLUE) approach to reconstruct a 3D magnetomotive displacement map, enabling improved visualization of the MNP distribution.

Fluorescence lifetime imaging ophthalmoscopy (FLIO) is a noninvasive imaging technique that allows in vivo measurement of autofluorescence intensity decays of endogenous fluorophores in the ocular fundus. So far, only devices from Heidelberg Engineering based on the Spectralis system, regarded as the gold standard, have been used in FLIO research. Here, we characterize a new FLIO device based on the RETImap system from Roland Consult and compare it with the gold standard using cuvette measurements. We determined the maximum laser power of the new FLIO device and analyzed the instrument’s behavior at three different laser power levels (150 μW, 200 μW and max.) in terms of laser spectrum and instrument response function (IRF). Then, alternating between the gold standard and the new device, we performed 80 FLIO measurements using fluorescent dyes filled in flat cuvettes (110-OS, Hellma GmbH & Co. KG, Müllheim, Germany), 20 measurements for each of the four following dyes: A (25 μM eosin Y solution mixed with 5 M potassium iodide solution), B (20 μM erythrosine B with water), C (15 μM eosin Y with 0.5 M potassium iodide), and D (3.3 μM eosin Y with water). For both devices and all measurements, the laser power was set to 101 μW. A rectangular ROI covering 1000 px was defined for each measurement. The mean fluorescence lifetimes of the ROIs were compared for the four dyes in the two spectral channels, short (498-560 nm, SSC) and long (560-720 nm, LSC). For the statistical analysis Shapiro-Wilk, Mann-Whitney U, F, and t-tests were used. Bonferroni correction was used to correct for multiple comparisons. The IRF and the full width at half maximum (FWHM) calculation and monoexponential fluorescence lifetime approximation were performed using the FLIMX software. The maximum laser power of the new device was 280 μW. For the consecutive laser powers 150 μW, 200 μW, and 280 μW, the IRF FWHM in SSC were 298.6±1.1 ps, 341.0±2.5 ps, and 347.5±6.0 ps, respectively. In LSC, the IRF FWHM were 290.4±3.8 ps, 344.0±3.4 ps, and 358.8±1.3 ps, respectively. All results are mean fluorescence lifetime ± standard deviation. Results of the comparison measurements showed statistically significant differences in mean fluorescence lifetimes between the two FLIO devices in SSC for dye D and in LSC for all fluorescent dyes. There were no statistically significant differences in mean lifetimes in SSC for dyes A, B, and C. A new fluorescence lifetime imaging ophthalmoscope has been characterized and is suitable for fluorescence lifetime studies. A comparison of the new device to the gold standard revealed that the devices differ significantly in the LSC and slightly in the SSC. Obtaining incompatible results in more than 60% of the cases tested implies that the results provided by the new device are not compatible with those obtained from the gold standard. It is planned to perform such a comparison on a larger number of measurements.

Undetected fluid-filled ovarian cysts can lead to severe complications like rupture and hemorrhage, requiring emergency medical interventions. Ultrasound is a commonly used imaging modality for detecting cysts and tumors. However, it presents challenges with weak contrast, blurred boundaries, and noise affecting the diagnostic reliability. Segment Anything Model (SAM) is a powerful framework for image segmentation, but struggles in medical imaging tasks due to its lack of domain-specific knowledge. MedSAM extends SAM to clinical data and demonstrates superior capabilities across various adnexal regions and imaging modalities. It lacks specificity for ultrasound images and underperforms in ovarian tumor segmentation. This limitation is addressed by efficient adaptation of MedSAM using attention-based adapter modules. Instead of fine-tuning the entire MedSAM, incorporating an adapter module is proven to be an efficient training strategy. The proposed approach optimizes model training using an attention based adapter, guided by a binary cross-entropy loss function with logits to ensure numerical stability. The adapted model achieves a dice score of 70%, demonstrating robust segmentation capabilities. With its ability to produce sharp boundaries and reliably segment regions of interest, this work highlights the potential of an attention-driven discriminative learning framework for ovarian tumor detection, contributing towards automated and effective computer-aided diagnosis of ovarian cancer.

Laser osteotomy enables precise and minimally invasive bone cutting but lacks real-time feedback on tissue type, increasing the risk of unintended damage to surrounding soft tissues such as tendons, muscles, nerves, and blood vessels. Accidental damage to these structures can compromise surgical outcomes and prolong recovery. To address this, we investigate the use of Laser-Induced Breakdown Spectroscopy (LIBS) for intraoperative tissue differentiation. While conventional LIBS methods rely on elemental intensity ratios, they often struggle to distinguish between tissues with similar compositions. In this study, we propose laser-produced plasma temperature as an additional stable parameter for differentiating hard and soft tissues. Experiments conducted on pig knuckle samples, selected for their anatomical similarity to human tissue, revealed clear plasma temperature-based separation between soft and hard tissues. These findings highlight plasma temperature as a valuable diagnostic parameter in LIBS-enhancing feedbackcontrolled laser surgery for better precision and tissue preservation, while also serving as a reliable input for deep learning models to enhance tissue classification accuracy. We foresee that the tissue differentiation process could be supported by temperature estimation via Boltzman plot of elements similarly in between some soft tissues.

In vivo confocal laser-scanning microscopy (CLSM) of the cornea provides image data from the corneal tissues with cellular resoultion. The ability to resolve the nerve fiber bundles of the sub-basal nerve plexus (SNP), in particular, opens the opportunity to use the SNP morphometry as a readily accessible, image-based quantitative biomarker for various neuropathic diseases. Several techniques have been proposed to overcome the small field of view of CLSM systems by capturing a dataset of many partially overlapping images covering an extended SNP area and creating a mosaic image from this dataset. So far, these mosaicking techniques have only been assessed qualitatively, by subjective visual inspection of the results. The present contribution describes an objective, quantitative evaluation strategy for these CLSM image montaging approaches by comparing the mosaicking results against ground truth data. Both the ground truth data and experimental datasets are acquired in a controlled experimental setup. Based on datasets from different specimens and different motion patterns, we benchmark our own mosaicking algorithm and report an "average of geometric distance" error value of ≤4.81 pixels. We provide the experimental data so that other researchers can evaluate and compare their own algorithms using the same base data.

Aortic valve (AV) stenosis is the most treated heart valve disease and the only curative treatment is the surgical AV replacement or the transcatheter prosthesis implantation. Bovine and porcine pericardium is used for fabricating biological heart valve prostheses but the durability is limited to 10-15 years. We use optical coherence tomography (OCT), a volumetric imaging technique with microscopic resolution, to monitor fiber structure and calcification of pericardium samples. Validated by histology, OCT at 880 and 1300 nm central wavelength captures the relevant structural features and shows a high potential for material testing of pericardium towards improved AV prostheses.

Mueller Matrix polarimetry (MMP) characterizes changes in light polarization after interacting with a medium, providing insights into tissue microstructure. Combined with multispectral (MS) imaging cameras, MS-MMP offers a novel way to quickly and safely acquire tissue surface information. Machine learning methodologies enable new diagnostic methods by automating tasks on fresh tissue biopsies, though this requires extensive and diverse data. To achieve this, we propose a user-friendly MS-MMP imager with a simple interface and fast acquisition time operated by laboratory technicians and residents. We show that our system, when operated by laboratory staff over several months, yields highquality data in large amounts and with positive feedback of its inclusion in a clinically compliant workflow. This positive outcome is promising for such systems to be used for large data collection initiatives.

Further knowledge of the regional pulmonary air distribution would deliver additional information of respiratory mechanics and could potentially improve lung therapy and oxygen supply. An electrical impedance tomograph and a spirometer were used to investigate whether regional air distribution in the lungs can be influenced by aimed breathing. It was examined whether four lung healthy volunteers were able to influence their breathing and breathe into specific parts of the lungs. The results showed that the air distribution within the lungs can be influenced in certain limits, even in untrained subjects. This indicates that improvements in lung therapy (e.g. drug treatment) might be influenced by trained patients.

Microtomography enables the three-dimensional imaging of cellular details of the entire mouse choroid plexus (ChP). This anatomical structure is vital for regulating the fluid balance in the brain. During aging, its epithelial cells flatten while its stroma expands. Quantification of the thickness of the ChP epithelial cell layer is hampered by the folded nature of the layers. We have developed a method to automatically quantify the thickness of the ChP epithelial cell layer even in the presence of folds and cell clusters. Visual inspection and tests with a wide range of parameter settings showed that the method is robust and provides realistic results.

Respiratory tract infections represent a significant global health burden. We developed a gold nanoparticle-based breath test device that enables rapid detection of infections at the point-of-care. The device measures the colorimetric properties of gold nanoparticles, which change upon aggregation when exposed to specific antigens filtered from exhaled breath. The system integrates sampling and analysis in a portable unit, providing results within 30 seconds. In this proof-ofconcept, we demonstrate the detection of SARS-CoV-2 antigens. Initial tests using NaCl-induced aggregation validate the device's functionality. The tests show the suitability for rapid, user-friendly infection diagnostics from exhaled breath. Further optimization is needed to enhance sensitivity and reliability.

In this paper, thin film magnetoelectric resonators are used to underpin an extension to the (modified) Butterworth-van Dyke (mBvD) model. These resonator types are being investigated for applications in the bio-medical domain for e.g., movement analysis or localization by sensing magnetic fields. The mBvD model provides an electrical equivalent circuit for such resonators, enabling the analysis of their response to applied currents or voltages. However, the resonators studied here change their properties - specifically, their resonance frequency - when exposed to a magnetic field. To better understand this behavior and adapt the resonator sensors to (medical) applications, the mBvD model is extended to include this property. Thus, to reflect the changes in the mBvD model, a dependency between the electrical components and the surrounding magnetic field is introduced. It is shown, that 4th order polynomial functions are suitable to map the changes and provide a robust model.

Optimizing sensor readout schemes and integrated circuit designs for both open-loop and closed-loop implementations requires precise modeling and simulation strategies. This study introduces a novel two-port impedance model to estimate the behavior of a converse Magnetoelectric (cME) sensor. This model provides a possible framework for calculating transfer functions and simulating magnetometer behavior in both continuous- and discrete-time simulation environments, and it is also possibly transferable to other magnetometer types. Common S-parameters were measured experimentally using an impedance analyzer and converted to Z-parameters to create a transfer function for system-level simulations. The model was validated through an analysis of output-related noise using MATLAB and LTSpice simulations to optimize the noise of the analog circuit parts of the system. The simulation results were compared with experimental measurements using a Zurich Instruments lock-in amplifier and the custom-designed low-noise printed circuit board (PCB) under model considerations. The proposed methodology derives noise considerations and the transfer function of a magnetometer. These are essential for readout schemes for mixed-signal circuit design. This allows low-noise electronics to be designed and extended to other sensor interface electronics, broadening their applicability in high-performance magnetic sensing.

Corneal diseases or injuries often result in vision loss, necessitating corneal transplantation. Donor corneas must meet strict quality standards, including sterility. To reduce annual corneal losses due to unintended contaminations, disinfection attempts using 222 nm irradiation were conducted and compared to 254 nm exposure. 254 nm is known for its antimicrobial impact but also for its danger to human cells, including corneal endothelial cells, while 222 nm is considered safe for human tissue, though it also exhibits high effectiveness against most microorganisms. In a laboratory study, porcine corneas were irradiated on the endothelial side with 150 mJ/cm² at 222 nm and 254 nm. Endothelial cells were counted after incubation in culture medium, salt solution, or demineralized water. The results were statistically analyzed and compared. For incubation periods up to 24 hours in culture medium, no significant differences (p = 0.715) between the applied wavelengths were observed. However, when demineralized water was used, both wavelengths induced cellular damage and a decline in cell density. Control samples revealed no detectable changes. The experiments demonstrated that porcine endothelial cells respond differently to the stress factor demineralized water compared to human endothelial cells. Additionally, photonic exposure combined with sufficient stress led to a reduction in cell density. However, it remains unclear how cell density would develop over several days without additional stress or whether a higher dose at 254 nm would yield different outcomes.

The increasing use of single-use flexible endoscopes raises concerns about their environmental impact. This systematic review includes 18 studies comparing singleuse and reusable endoscopes based on quantitative analyses on their environmental impact. Types included are bronchoscopes, cystoscopes, duodenoscopes, gastroscopes, laryngoscopes, and ureteroscopes. The number of available studies differs by endoscope type, with cystoscopes and bronchoscopes being most frequently assessed (seven and four studies). Findings indicate that the environmental impact depends on the specific endoscope type, with single-use cystoscopes being favorable, while for duodenoscopes, data suggest the reusable option to be more sustainable. However, results are highly study-dependent, influenced by regional circumstances such as transport distances, waste management, and reprocessing assumptions. Further quantitative data from underrepresented regions could help close remaining knowledge gaps for gastroscopes and laryngoscopes, and innovative materialsaving technologies may help reduce the carbon footprint of disposable devices where indispensable.

Single-channel SQUID system technology, operating at a noise level of 100’s of aT/ √ Hz, enables the non-invasive detection of synchronized spiking activity at the single-trial level via magnetoencephalography (MEG). However, when combined with simultaneous electroencephalography (EEG) recordings, the noise performance of the ultrasensitive MEG system can be greatly diminished. This issue negates some of the complementary qualities of these two recording methods. In addition, typical electrical components required for electrical stimulation of peripheral nerves, a common method for evoking specific brain responses, are also observed to have a detrimental influence on ultra-low MEG noise performance. These effects are caused by electromagnetic interference (EMI) and typically preclude single-trial detection. This work outlines, how careful design allows a significant reduction of the impact of EMI when these different electronic systems are operated concurrently. This optimization enabled the simultaneous single-trial detection of synchronized spiking activity using these two highly sensitive recording modalities.

Hand-Arm Vibration Syndrome (HAVS) is a significant occupational health concern, particularly among surgeons who use power tools like power-assisted liposuction (PAL) devices. In this context, this study evaluates the vibration exposure of a commercially available PAL system by using a 3D-printed finger adapter equipped with a affordable tri-axial accelerometer. Measurements were conducted on the handpiece alone and with a cannula attached performed by three operators. Vibration exposure was calculated according to DIN EN ISO 5349-1 standards. Fast Fourier Transform (FFT) analysis confirmed a primary vibration frequency of 75 Hz, consistent with prior research, thus validating the adapter’s reliability in capturing the frequency spectrum. The mean overall vibration magnitude (ahv) was 0.54 +-} 0.05 m s2 for the handpiece alone and 2.78 +-} 0.1 m s2 with the cannula attached. The observed precision, indicated by a small standard deviation, suggests the adapter’s consistency under controlled conditions. While our measured vibration magnitude was lower than the manufacturer’s reported value (3.77 m s2 ), this difference likely reflects the manufacturer’s testing of larger cannulas. The findings highlight the promise of utilizing a 3Dprinted finger adapter, coupled with an low-cost accelerometer, as a accessible and consistent method for quantifying PAL device vibration. Further research is recommended to validate its accuracy across various surgical settings and devices, and to explore its utility in developing effective HAVS mitigation strategies.

Sonodynamic therapy, which combines lowintensity ultrasound with sonosensitizers, shows promising anti-cancer effects, yet its underlying mechanisms remain incompletely understood. To support systematic in vitro studies, we developed and characterized an ultrasound setup designed to expose cells to well-defined and reproducible conditions. The setup includes a water tank, a sample holder, and absorbing structures to minimize stationary waves. Hydrophone measurements were conducted to assess the ultrasound pressure field, and finite element method (FEM) simulations were used to complement these measurements. The simulations allow for insights into power and intensity distributions that are challenging to measure directly. Results showed good agreement between measured and simulated pressures, although discrepancies of up to 20% were observed in ring patterns present in the simulation, but absent in the measurements. The study highlights the importance of simulations for optimizing experimental design and interpreting measurement artifacts, particularly those created by the hydrophone itself.

In order to integrate highly sensitive magnetometers based on surface acoustic wave (SAW) sensors into clinical applications, some fundamental challenges regarding their readout must first be overcome. The present paper proposes a readout concept that is based on a phase-locked loop (PLL) to demodulate a frequency-modulated (FM) signal. The presented demodulator is designed to meet three fundamental requirements: it is able to tolerate the frequency drift of the sensor system, it is able to demodulate low frequency as well as low amplitude and broadband signals, and it is sufficiently inexpensive and simple to be scaled for arrays. Preliminary measurements demonstrate that the chosen topology meets these requirements to a satisfactory extent, indicating its potential for broader implementation in clinical settings.

Accurate estimation of left ventricular volume (LVV) is essential for managing cardiovascular diseases such as heart failure and myocardial infarction. A promising approach for continuous LVV estimation is using intracardiac admittance measurements. The prevalent methods in literature are based on analytical models, for which several simplifying assumptions had to be made. It is not clear if the models are the optimal choice to describe the relationship between the admittance measurement and the LVV. This study explores intracardiac admittance measurements as a promising alternative for continuous LVV estimation. We utilized symbolic regression to derive mathematical models describing the relationship between intracardiac admittance measurements and LVV without predefined structures and compare their performance with classical models from literature. Through simulations based on a cylindrical left ventricle model, we rediscovered the model of Baan and identified further interpretable models that outperform the approaches by Baan and Wei on our dataset. Our findings support that symbolic regression can uncover meaningful patterns in biomedical data, paving the way for more efficient diagnostic tools in cardiovascular health monitoring.

We present a custom fluid phantom usable in the development of sensing technologies to assess the pulse wave. The phantom provides adjustable flow and pressure to an interchangeable target volume and integrates sensors to provide reference readings on fluid dynamics. The proposed system has a modular design and supports the evaluation of various sensing technologies. In this paper, we describe the phantom and demonstrate its basic functionality using the integrated sensors and considering a camera for imaging photoplethysmography as a device under test. Our experiments show reasonable results regarding the performance of the phantom and highlight opportunities for its future use. Due to a wide applicability, easy expansion options and a cost-effective design, we believe that the phantom can contribute to future research activities, not only by our group but even by other groups.

Huge parameter studies are usually conducted to optimize and characterize systems for steering magnetic nanoparticles in magnetic drug targeting (MDT). Thus, many simulations or measurements are needed to determine the system which is time-consuming and expensive. To overcome this problem, this paper investigates the feasibility of using orthogonal arrays according to Taguchi’s method for determining parameter trends in MDT. These trends can then be used to further optimize the appropriate parameters. By doing so, parameters typically varied in the literature were changed in three levels in a simulation and measurement scenario. In these scenarios, using orthogonal Taguchi arrays reduced the number of analyzed parameter combinations from 243 to 9. Overall, the results reveal that the relative importance of the parameters was correctly identified in the simulation and measurement scenario. The trends of the parameters showed some outliers due to single dominant parameters or correlations among them. However, this is normal for the analysis with orthogonal Taguchi arrays. Thus, it can be concluded that using orthogonal Taguchi arrays to optimize and characterize steering systems in MDT dramatically reduces the optimization time.

Detection of specific biomarkers in blood plasma is essential for fast and accurate diagnosis and treatment of medical conditions. Hand-powered centrifuges offer a promising alternative to conventional centrifuges for blood seperation in resource-limited settings. Though these systems are efficient for blood plasma extraction, the necessary steps for performing the analysis remain. In this paper a concept for automatic combining of a detection reagent with blood plasma during hand-powered centrifugation is presented. Evaluation showed the successful routing of a reagent carrier to a defined position after the designed number of centriufagtion cycles. Furthermore it could be shown, that even with varying applied force (exceeding a minimum value) the extracted plasma volume is sufficient when combining with the reagent. The system reached a maximum rotational velocity of up to 9000 rpm which has proved to be equivalent to about 4500 rpm on a conventional centrifuge.

Aim and focus of the research was to develop a prototype system to increase training intensity of the paretic arm and functional cognition for stroke patients through homebased rehabilitation. The key feature of this system is its ability to deliver home-based dual-task training with haptic feedback for grasp, addressing the high priority of grasp function in upper extremity rehabilitation. The prototype consists of a gamified training application and specifically developed input devices. It is the first sensor-based rehabilitation system to incorporate neutral hand positioning, increasing comfort during use and offering physiological benefits. Real-time feedback on cognitive and motor performance is expected to encourage patient engagement. The prototype received positive feedback from experts regarding its potential effectiveness in a preliminary evaluation with seven participants. In future research, a pilot study and real-world testing with the target group will be necessary to objectively verify whether the use of the prototype increases training intensity.

Patients depending on home parenteral nutrition (HPN) must follow a strict procedure for catheter care, as insufficient catheter flushing and hygiene can lead to health risks, including catheter occlusion and infection. Currently, flushing procedures rely on best practice based on experiences passed on from healthcare professional to healthcare professional and no standardized guidelines are given to ensure effectiveness. To address this, this study introduces an experimental setup to evaluate different flushing processes under controlled conditions. The setup uses a peristaltic pump, that generates flushing patterns with adjustable flow rate, pulsing frequency, and volume. A flow meter records the flushing characteristics for analysis and is used to control the exact volume of flushing liquid used. Each flushing method is assessed by measuring the removal of contamination from the port cannula. Unlike prior studies, that focus on either fat or protein emulsions, this research considers the complete nutrient solution. This method provides a reliable and reproducible way to evaluate the flushing process and could be used to identify an optimal flushing procedure. Subsequently, this optimal flushing procedure can form the basis for the development of an automatic flushing system that can be integrated into the overall automation process, using only one pump for both parenteral nutrition and flushing of the port catheter. Finding an optimal flushing procedure and developing an automated flushing process can support the standardization in port catheter care, improve patient autonomy and safety and reduce dependency on healthcare professionals for HPN patients.

Pictograms are commonly used to document probe positions in diagnostical ultrasound procedures. A new computer-vision based method using depth cameras is proposed and tested to automatically document the 3D probe position relative to the patient body. The probe is held in different poses by the examiner and translation magnitudes and tilting angles are compared to reference values obtained by a robotic arm. Our results express the systems general suitability for visual documentation. Further optimization of the processing pipeline for this use case is planned, followed by a study in a clinical environment.

Dehydration can cause serious health problems. The optimal amount and the right time to drink varies individually and depends on many factors, i.e. anthropometric data, current stress levels or environmental influences. Therefore, we aim to develop an intelligent hydration management system for sports and health care applications based on sensor technology to determine fluid intake, linked to current vital, performance and environmental parameters. This allows the calculation of individual requirements and drinking recommendations for an optimal fluid and electrolyte balance. In the current study, we evaluated three sensor principles for determination of drinking volumes and analyzed boundary conditions for an automated test stand to simulate tube-based drinking.

Atrial fibrillation (AF), the most common sustained arrhythmia, affects over 59 million people worldwide and carries a high risk of serious complications such as stroke. However, AF often goes undiagnosed, leaving many patients untreated. Photoplethysmography (PPG) is a non-invasive method for AF detection, accessible via wearables or smartphone apps. This paper presents two integrated smartphone apps-one for user registration and one for PPGbased AF screening-developed for use in a large-scale, prospective, randomised, centreless clinical trial. The screening app incorporates third-party AF detection Software Development Kits (SDKs) which are certified as class IIa medical devices. A pilot study confirmed the technical feasibility of the approach and identified minor areas for improvement. These insights will guide app refinement ahead of a nationwide AF screening study in Austria, planned for 2026.

The entoptic view of the retina is a fascinating phenomenon that is caused by light stimulation. Over the course of the past two years, we have been developing our Purkinje Retinal Image Stabilizer (PRIS) method to simplify entoptic testing for retinal examinations. Our non-invasive method is intended as a retinal function test and to make disease-related changes in the retina or the retinal blood vessels visible to the user. Technically, we have implemented it in a small light stimulator, the PRIS-Tool. In a next step, clinical studies have to be conducted to verify the reliability and safety of the PRIS method as an ophthalmic screening tool for medical applications. We are focused on its use for the early detection of diabetic retinopathy worldwide and for the preoperative assessment of retinal function in cataract patients in low-income countries. For clinical studies, enhancements and new functions have been implemented in the redesigned PRIS-Tool device to facilitate handling and optimize in-house assembly. We also present our approach for assessing photobiological safety in accordance with IEC/EN 62471.

Accurate localization of arrhythmogenic substrates is crucial for effective stereotactic arrhythmia radioablation (STAR) in ventricular tachycardia (VT) patients. Conventional CT registration methods for left ventricular (LV) strain estimation face challenges due to image quality limitations and motion coupling. We propose a novel modelbased CT registration algorithm that incorporates anatomical constraints to enhance LV strain estimation. Our approach segments the LV myocardium using 3D Slicer’s TotalSegmentator extension, models it as a harmonic scalar field, and discretizes it into multiple layers to preserve layerspecific motion. A non-rigid registration technique then aligns end-diastolic and end-systolic volumes to compute regional longitudinal strain (RLS). Evaluation on CT data from ten VT patients demonstrates superior alignment and anatomical accuracy compared to conventional methods, potentially improving LV strain estimation for targeted therapy.

Relative movements between electrodes and tissue are a potential source of interference signals that can affect the accuracy of signal recordings and data analyses. To address this challenge, a test bench is developed that utilises mechanical excitation to impose dynamic flow on the electrodes under various conditions to characterize the influence on signal quality. The generated data can then be recorded and analysed with the Fast Fourier Transformation.

Despite the large variety of traditional, flexible and stretchable Micro-Electrode Array (MEA) devices available for use with organoids, most of them present drawbacks when applied for long-term 3D neural tissue monitoring applications. In many cases, 3D neural tissues can only be monitored for acute and/or short-term durations due to tissue necrosis that occurs due to the lack of tissue oxygenation under submerged culture conditions. This paper presents alternative MEA devices, called ALI-MEA. These ALI-MEA allow the maintenance of 3D neural tissues at air-liquid interface, which provides excellent tissue oxygenation and long-term survival conditions, resulting in the capability to stimulate and record electrophysiological activity from the tissues for long-lasting experiments, up to several months. This paper illustrates these findings with results from several typical 3D neural tissue preparation experiments showing that long-term experiments can be easily achieved using the ALI-MEA devices.

Cardiotoxic events are a major challenge in drug development and a leading cause for market withdrawal. Cardiotoxicity frequently originates from electrophysiological alterations within cardiomyocytes, therefore increasing risk of life-threatening arrhythmias. Current preclinical 2D cell culture models lack the structural and functional complexity of mature myocardium. To address this limitation, we validated a microphysiological system (MPS) with porcine ex vivo living myocardial slices (LMS) incorporating a dualpacing strategy (S1S2) that was developed for direct assessment of effective refractory period (ERP) duration of action potential. Ex vivo cultivated porcine LMS responded to dual pacing and displayed action potential durations comparable to healthy human hearts. Specific human ether-ago- go-related gene (hERG) potassium channel blockade with Dofetilide prolonged ERP in a dose-dependent manner. Of great interest, our model was validated by mimicking the clinically cardiotoxic side effect of Cisapride. Titration of Cisapride triggered an elongated ERP, thus confirming cardiotoxicity screening power. Our newly developed MPSlms Tox platform enhances translatability by incorporating mature cardiac tissue and potently allows multidose drug testing from a single donor heart thereby elegantly reducing animal studies.

Valvular heart disease (VHD) is the third most common cause of death in the world and constitutes a significant concern for global health. Traditional diagnostic methods, such as echocardiography, computed tomography, and magnetic resonance imaging, are effective, but their limitations in outpatient monitoring have led to exploration of alternative techniques based on wearable inertial measurement unit (IMU) sensors. This study examined differences in Hjorth parameters (activity, mobility, and complexity) between 29 healthy individuals and 30 patients with VHD using electrocardiograms (ECG signals), seismocardiograms (SCG signals), and gyrocardiograms (GCG signals) in healthy subjects from two public datasets: “Mechanocardiograms with ECG reference” dataset and “An Open-access Database for the Evaluation of Cardio-mechanical Signals from Patients with Valved Heart Diseases”, respectively. No significant differences in complexity for all signals and activity in SCG signals were observed. Statistically significant differences (twosample t-test, p<0.05) in mobility for ECG signals and in activity for GCG signal highlight their potential in distinguishing VHD patients from healthy individuals. The study supports the potential of signal-based markers for wearable cardiac monitoring. Future research should focus on larger and more diverse datasets, machine learning-based classification, and clinical integration to improve cardiac diagnostics.

Pressure ulcers pose a threat, especially for older people and ICU patients. Data-driven approaches could allow for early detection and continuous evaluation. In this work, a deep convolutional generative adversarial network is presented which is able to generate new images of pressure ulcers. The cavities concerning the training dataset and the transferlearning strategy are analyzed to optimally supplement small clinical datasets for more robust algorithm development.

This study investigates the contribution of speech audio and speech verbal content in the automated detection of depression levels. Recordings from the Distress Analysis Interview Corpus Wizard-of-Oz dataset and the depression severity labels of the recordings were used to extract acoustic features. A transcription of the recordings was used to extract textual features. The acoustic set included prosodic, cepstral, and glottal feature categories. The textual features consisted of semantic and syntactic categories. Mutual information feature selection, followed by a random forest classifier identified the set of features which optimised the depression level classification. The optimised binary classification of depression from non-depressed yielded an accuracy of 0.89 and an F1 score of 0.87. A classification of the five depression levels yielded an accuracy of 0.79 and an F1 score of 0.72. The ratio of importance scores of acoustic to textual of the speech acoustic features was greater than 3:1. Our method thus provided acoustic and textual indicators in depressed speech. These might increase the acceptability of automated depression detection by healthcare professionals. Our initial findings indicate a select set of features that can improve the effectiveness of automated depression detection and monitoring tools.

Cardiovascular disease is a global concern with a high economic burden that needs to be controlled. Predictors of cardiovascular disease include blood pressure. There is a drive to evaluate these predictors through continuous non-invasive methods, which often requires custom devices to achieve the measurements. To reduce reliance on custom devices, we developed, verified, and validated an algorithm to measure the dynamic cross-sectional area of the radial artery and estimate the pulse rate from brightness-mode ultrasound imaging. The algorithm was implemented using Python and MATLABR, and verified using a simulated environment with known parameters. It was validated using radial artery ultrasound scans, where pulse rate estimation was confirmed using continuous non-invasive arterial pressure monitoring as the control. The algorithm was verified to measure the area at >99%accuracy with a difference in pixel value between artery and background >20 when considering a complex signal. The pulse rate estimate was exact. The accuracy of area measurement decreased when applied to ultrasound scans. However, the algorithm still estimated the pulse rate within the error margin, compared to the control, for five out of six subjects thus validating the algorithm. This demonstrates the possibility of measuring physiological factors without the need of custom devices.With further development, this algorithm could incorporate colour doppler information such that arterial pressure could be determined using a medical ultrasound device.

Two- or three- dimensional point clouds are common representations of data in the medical context, especially in medical imaging and segmentation. When assigning realvalued coordinates and transformations - e.g., for registration with an additional imaging modality - metrics like the Dice similarity coefficient, which rely on set operations on discrete point sets, have to be approximated. We applied different calculation methods for multiple metrics on a synthetic dataset of basic shapes, combined with modifications like transformations or occlusions. The evaluation resulted in a wide range of values for all metrics, depending on the modification, calculation method, and parametrization. This variability suggests both a potential clinical risk and the opportunity for selectively reporting more favourable results in publications. Therefore, and considering the influence of point spacing and distribution, we recommend reporting multiple metrics from diverse categories in the context of real-valued point clouds.

Implant-associated infections pose a significant challenge in modern medicine, particularly due to the formation of biofilms that lead to a high tolerance of the bacteria against antimicrobial treatments [1]. This study investigates an alternative approach for the prevention and treatment of biofilms on implants and adjacent bone tissue by targeted blue light irradiation with a wavelength of 405 nm. The irradiation experiments were carried out with biofilms of Escherichia coli (E. coli) cultivated on bovine bone samples. The effectiveness of the irradiation was investigated both regarding the biofilm using crystal violet (CV) staining as well as specifically against the embedded bacteria by determining the colony forming units (CFU). The results suggest that irradiation with 405 nm at an irradiation dose of 5.4 J/cm2 initially leads to no significant reduction in biofilm mass. After irradiation with a dose of 10.8 J/cm2, a significant biofilm reduction is observed. However, after prolonged irradiation yielding an irradiation dose of 54 J/cm2, renewed biofilm formation is detected, indicating a limited penetration depth of the light and possibly adaptive mechanisms of the bacteria. The data reveal that while the irradiation with 405 nm light with the applied radiant exposure was able to partly eradicate biofilms, it does not guarantee reliable eradication of the embedded bacteria. Nevertheless, this method could attain more effective results when applied with higher radiant exposures or in combination with existing treatment approaches.

Manual annotation of biomedical data is often a prolonged and error-prone process, particularly in scenarios involving dynamic physical interactions, such as needle insertions into soft tissue. In this study, we present an automated annotation framework using the Segment Anything Model (SAM) to detect puncture events in vibroacoustic recordings from needle insertions into preserved Manduca sexta specimens. The annotation tool utilizes SAM to extract segmentation masks from video recordings of needle insertions and analyzes vertical deformations of the resulting mask boundaries. Two puncture moments, one at entry and one at exit, were estimated by searching for minima in the bottom boundary displacements, yielding more reliable results compared to the top boundary. The tool detected the first and second punctures in 85% and 90% out of 300 cases, respectively. Although the tool provides promising accuracy, further refinement could be done to achieve higher and more robust precision. Limitations like background interference, reflections, and needle inclusion can be mitigated by improving the setup or optimizing segmentation models, including fine-tuning SAM.

Dynamic fatigue testing of medical devices represents a complex and expensive process. Universal electrodynamic testing machines are costly. Price scales with the number of devices under test. For accurate testing, the machine must produce the specified conditions. This includes amplitude and frequency but also the loading curve shape. While continuous tracking of amplitude and frequency has become standard practice, current methodologies still rely on optical and intermittent observations for analyzing the curve shape. In this study, we present a solution that combines cost-effective electrodynamic fatigue testing devices with a custom feedback control system. The system uses a windowed discrete Fourier transform with peak finding to determine the relevant parameters. The proposed system has been deployed in two test setups. A transfer to further test applications can easily be implemented.

This paper proposes a novel approach for the noninvasive assessment of different degrees of stenosis by remote photoplethysmography (rPPG) using an RGB camera. A proof-of-concept study was performed on 20 healthy participants simulating different levels of perfusion in the upper extremities. The rPPG signals from both arms were extracted and the pulse transit time (PTT) was analyzed using cross correlation. The results show that the PTT time can be used for both the detection of stenosis and the differentiation of two degrees of stenosis. These findings have the potential to contribute to the development of a low-cost and non-invasive diagnostic approach for peripheral arterial disease (PAD).

Cardiovascular diseases (CVDs) remain the leading cause of mortality worldwide. New sensor-based approaches to assess cardiovascular risk may support individualized healthcare. Recent advancements in Artificial Intelligence (AI) suggest Electrocardiogram (ECG) - based predictions of age and sex as potential biomarkers for cardiovascular risk stratification. Discrepancies between predicted and chronological age, as well as inconsistencies in sex classification, have been linked to increased cardiovascular risk. However, these data-driven approaches also contain the risk of not equally performing across subgroups. In this work, we introduce a deep learning model that predicts both ECG-derived age and sex simultaneously. Trained on the CODE dataset and validated on PTB-XL, the model achieves a mean absolute error of 8.85 years in age prediction and an Area under ROC of 0.93 for sex classification. Notably, sex prediction accuracy decreases with age, suggesting morphological changes in ECG signals. To enhance interpretability, we use posthoc explainable AI analysis to highlight ECG regions relevant for sex classification, revealing known sex-specific features. Our findings underline the potential of AI-driven ECG analysis as a promising, explainable tool for cardiovascular risk assessment.

Hand gesture recognition is an important task in human-machine interaction, enabling intuitive and accessible control methods across various applications, including assistive technologies, virtual reality or sign language translation. The gesture recognition task relies on sensory input from different modalities and in order to be processed, the data needs to be labeled with the corresponding hand gesture classes. This is often done manually which remains a major bottleneck in developing robust gesture recognition systems. This paper presents an Auto-Labeling Pipeline designed to automate the annotation process for multi-modal sensor data, reducing the time and effort required for labeling input data. The proposed system uses data from the Ultraleap Leap Motion Controller, and classifies predefined gesture models based on inter-joint angles and the cosine similarity between their pose vectors. The pipeline proposes a post-processing step to filter misclassifications and enhance gesture recognition reliability. The developed open-source toolbox enables researchers to collect and label gesture datasets efficiently, making hand gesture recognition more accessible and scalable.

Patients with pacemakers represent a high-risk group for cardiovascular events. While deep learning has recently shown promising results in detecting many abnormalities in electrocardiograms (ECGs), little attention has been paid to its performance in the presence of pacemaker activity. In this study, we evaluate the impact of pacemakers on the performance of a state-of-the-art deep neural network (DNN). Using the MIMIC-IV ECG dataset for validation, we compared model performance between ECGs from patients with and without implanted pacemakers. We observed a notable decrease in performance in pacemaker patients compared to non-paced patients (area under curve (AUC) 0.850 vs 0.770, sensitivity 0.630 vs. 0.423). To understand this discrepancy, we applied an explainable artificial intelligence (XAI) method to analyze the relevance of the model’s predictions. Lead V1 was identified as the most relevant lead for the prediction of the model, even in the presence of a pacemaker. In addition, the false negative (fn) predictions of the model for pacemaker ECGs were most influenced by the P wave segment in lead V1. Our findings highlight the need for careful adjustment and training of DNNs in healthcare to achieve fair models that generalize well across diverse patient populations.

Narrowing and stiffening of arteries are a major cause for cardiovascular diseases such as heart attacks and hypoxia. A current inflatable balloon sensor system integrated on a guide catheter system is being developed to diagnose the degree of narrowing in vessels and its shape and determine localized tissue stiffness. The local ‘tactile’ strains on the balloon segments are intended to be converted into predictive assessment and decision aids for the surgeon. This contribution in the first step forward simulates in 2D an ideal balloon inflation within different rigid shapes of equal perimeter and captures the local stretch information from the balloon segments and the enclosed area of the balloon in the conformed state. This information serves as an input of an inverse model to create to estimate the tissue shape. The inverse model utilizes the difference of the individual segment stretches and the mean of segment stretches to compute the external angle formed by the segments. The reconstruction of the basic simulated lumen profiles is obtained with fairly accurate metrics though some discrepancies about the reconstructed area and accuracy of actual shape arise in few simulations. The study provides some information on the measurement sensitivity and the reconstruction and suggests some other refinement and techniques may be useful for reliable and accurate reconstruction.

Passive Acoustic Mapping (PAM) is a technique used to localize cavitation events in biomedical applications such as targeted drug delivery, histotripsy, and lithotripsy. Accurate cavitation mapping is essential for optimizing therapeutic efficacy and safety. However, conventional PAM assumes a homogeneous speed of sound within tissue, which does not reflect the heterogeneous acoustic properties of biological media. In this work, we introduce the Fast Marching Method (FMM) to PAM for accurately computing time delays in arbitrary speed of sound distributions. We evaluate the proposed approach through simulations, demonstrating its potential for improved cavitation mapping in realistic tissue environments.

As materials such as Poly-Lactic Acid (PLA) have become a significant part of medical production and research, it is important that the resulting products adhere to strict regulations concerning contaminations. A wide variety of methods are currently used to comply with these regulations depending on the products and their intended use case. We propose a standardized workflow for high-sensitivity contamination detection in X-ray photoelectron spectroscopy (XPS) survey scans that helps enhance detection reliability, reduce throughput constraints associated with manual analysis and streamline an automated analysis. This makes XPS a more viable option for process monitoring and reduces the need for additional analysis steps. The workflow uses a neural network trained on synthetic signals to predict contamination at%. These are then compared with individually variable contamination limits to classify each contaminant.

Epilepsy represents a chronic non-communicable brain disease which affects about 70 million people worldwide. The condition presents with recurring seizures which produce brief periods of unmanageable movement. EEG serves as the primary diagnostic instrument for epilepsy diagnosis to date.We examined seizure EEG signals from Physionet database based on Hjorth parameters (Activity, Mobility, Complexity), k-Nearest Neighbors (k-NN), Random Forest, and Decision Tree. The results (between 0.9827 and 0.9999 of accuracy) demonstrate that Hjorth parameters can be used for detecting seizure episodes, despite class imbalance in the dataset. The best-performing approach was the hybrid system based on the Random Forest (1.0000 of accuracy).

Blood pressure (BP) is a key indicator for cardiac diseases. Although many current works focus on estimating BP using photoplethysmography (PPG), the number of works using imaging PPG, i.e. PPG derived from videos, and works to compare between different PPG settings is limited. This study compares BP estimation methods using regression ensembles across contact and non-contact PPG settings by means of absolute error (MAE), Pearson correlation coefficient and feature importance. Our results show similar BP estimation quality for both contact and non-contact variants. While MAEs are, similar to comparable works, high (≈ 13mmHg and 8mmHg for systolic BP (SBP) and diastolic BP (DBP)), particularly for DBP and pulse pressure our analyses yield promising correlation coefficients (> 0.7). We observed notable differences in feature importance underlining the relevance of acquisition setup and processing strategies. This work highlights the potential of iPPG for BP estimation, although further research is needed to fully understand the advantages and limitations of both contact and non-contact methods.

The application of instrument segmentation holds significant potential for the future of robotic-assisted minimally invasive surgery, as it has the ability to enhance context awareness of the support system. The objective of this research is the development of a new feature that detects the tips of an instrument and can measures the distance between them. The calculated distances between the tips can then be utilized by the surgeon to assess the geometric structure of organs or lesions, which can play a pivotal role in the decision-making process. The proposed methodology involves the automatic segmentation of instrument parts using Mask R-CNN, which facilitates the selection of a tool tip through a straightforward algorithm. The necessity of connecting jaws to the subsequent instrument part or the boundary is beneficial in the elimination of outliers. The Mask R-CNN achieves an mIoU of 0.6441 for instrument parts segmentation. The tips are detected with a rate of 66.78% compared to the ground truth. The average euclidean distance to the ground truth is 8.37 pixels or 1.02% of the image resolution.

Hyperspectral Imaging (HSI) is a promising tool for assisting medical diagnostics, as it enables precise tissue characterization and differentiation through detailed spectral analysis. However, HSI data analysis faces challenges in reliably identifying relevant regions (e.g., tumors, vascular structures) due to spectral variability, complicating universal algorithm development, especially with limited ground truth data. Focusing on endometriosis as a medical use case, we generate plausible reference spectra (endmembers) for affected tissue using HSI data, despite limited and weakly annotated datasets. Our processing pipeline includes Savitzky-Golay (SG) smoothing, Standard Normal Variate (SNV) standardization, Principal Component Analysis (PCA), superpixel segmentation and the Pixel Purity Index (PPI). Validation via Intersection over Union (IoU) achieves 0.94 accuracy in lesion detection.

Deep Brain Stimulation (DBS) is an effective treatment for movement disorders. Optimizing stimulation parameters remains, however, a trial-and-error process. Datadriven models leveraging Probabilistic Mapping have shown promise in predicting DBS outcomes, yet current studies rely on chronic stimulation data. This study explores the feasibility of using intra-operative stimulation test data for DBS effect prediction. Probabilistic volumes of beneficial and adverse effects were computed from intra-operative stimulation test data of 65 patients (23 with Essential Tremor + 42 with Parkinson’s Disease). A prediction dataset was generated including clinical, morphological, stimulation features along with features derived from probabilistic maps and simulated Volumes of Tissue Activated. Three machine learning models (Adaboost, Support Vector Classifier and Naïve Bayes) were implemented to predict stimulation effects in a classification task. The models were validated in a leave-one-out crossvalidation and their performances were compared. All the developed models were able to predict DBS outcome classes. The best predictive performance was achieved by the Adaboost model with a maximum balanced accuracy of 0.71 on 3 classes. These results show that intra-operative stimulation test data can predict DBS effects with a similar approach and comparable accuracy to post-operative monopolar review data.

The end-tidal partial pressure of CO 2 (PetCO 2 ) is widely used as a noninvasive and continuously measurable surrogate for the arterial partial pressure of CO 2 (PaCO 2 ). The arterial to end-tidal CO 2 gradient (ΔpCO 2 (a-et)), which is usually positive, changes over time and may become negative. Hence, estimating PaCO 2 from PetCO 2 is challenging. Yet, knowledge on the prevalence of negative ΔpCO 2 (aet) is limited. We analyze a dataset of 63 mechanically ventilated preterm lambs on the occurrence of negative ΔpCO 2 (a-et). Among 1,182 paired measurements of PaCO 2 and PetCO 2 , 185 instances of negativeΔpCO 2 (a-et) are found, which are distributed among 36 subjects. Further, we present three classifiers for detecting negative ΔpCO 2 (a-et) on a balanced subdataset based on noninvasive measurements of routine monitoring. The classifiers achieve promising results and when validated on human data can contribute to robustly estimating PaCO 2 .

Screening of cancer drugs in personalized medicine and cancer treatment research is expensive. Large libraries of compounds must be evaluated, and multiple doses for each compound need to be tested to assess their viability. We introduce a method exploiting cost-aware Gaussian process Bayesian optimization to reduce the number of experiments and amount of compound spent in screening. The method is utilized to iteratively find suitable medication doses, while staying within a predefined budget. Our approach is validated using synthetic data, and subsequently tested using experimental data.

Deep learning methods are increasingly used in clinical artificial intelligence (AI) research, including for detecting anomalies in intensive care data. However, their evaluation often depends on human annotations, which can vary in quality and consistency. In this study, we analyse the effect of annotation variability on the performance of DeepAnT, an unsupervised convolutional neural network for anomaly detection (AD). Using intensive care time-series data from 38 patients for training and six patients separately annotated for evaluation, we compare F1 scores based on two independent physician annotations. Our results show differences in model performance across different vital parameters, between patients, and especially between annotators evaluating the same data. These findings indicate that human labelling has a measurable impact on the perceived performance of the AD algorithm. Structured labelling protocols may be beneficial for achieving more consistent and reliable evaluations.

Glaucoma is a leading cause of irreversible blindness, requiring early detection to prevent vision loss. This study evaluates non-linear machine learning (ML) models, such as AdaBoost, for forecasting glaucoma stage, comparing it to Logistic Regression (LR) using visual field data from three independent cohorts (Bern, Rotterdam, Washington). AdaBoost outperformed LR, achieving a kappa value of 0.85 (Bern) compared to 0.73 for LR. Notably, AdaBoost reached an AUC ROC of 0.93 (early glaucoma, Bern), surpassing LR’s 0.47. Similar trends were observed across other datasets. These results demonstrate the ability of non-linear ML models to detect glaucoma progression, highlighting their potential to improve early intervention and treatment planning.

Detecting Ineffective Efforts during Expiration (IEE) in invasively ventilated neonates could help to improve the ventilation process and lower the discomfort of the patient. We propose an intrinsically explainable rule-based decision tree to detect IEE in high-frequency airway flow data. The algorithm is evaluated and compared to a previously developed Long Short-Term Memory (LSTM) neural network. The decision tree achieves a lower yet comparable performance to the LSTM, while being more stable during training and needing considerably less time for fitting and inferring. Further, the decision tree achieved good results on only 20% of the training data, which in future could reduce annotation labor. Overall, the results encourage us not to disregard rule-based algorithms as outdated, especially when explainability is of importance and annotated data is scarce and time-intensive to acquire.

Early detection of cognitive impairment is crucial for timely intervention in conditions like dementia and Parkinson’s disease. While the Mini-Cog is a validated tool for such screening, its traditional clinical format may disengage some users. This study introduces a gamified, narrative-based version of the Mini-Cog as a web-mobile application, tested with 30 healthy volunteers (ages 45-72) and one patient. Two story variants were developed, featuring a mayor and a dinosaur, embedded in a relaxed everyday narrative mentioning a park, bakery, and birthday party. Feedback highlights positive engagement through visuals and icons, with suggestions or age-adapted stories and improved screen adaptation. Results support the feasibility of gamified cognitive screening while noting the need for broader clinical validation.

The assessment of the efficacy of ventilators and sleep therapy devices requires either medical studies, whose reproducibility on patients is a challenge, or simulative bench tests using adequate hardware. The ASL5000 lung simulator (IngMar Medical, Pittsburg, USA) is one of the most widely utilized tools used by manufacturers of ventilatory devices and bench test groups for this purpose. This study systematically investigates for the first time its dynamic simulating accuracy in two operational modes: flow pump mode and lung model mode. Our results indicate that in flow pump mode, the ASL5000 exhibits resonance phenomena and substantial attenuation at higher frequencies. These characteristics may constrain the simulator’s accuracy, particularly when using test signals with rapid flow variations, such as flattening or snoring. In lung model mode, attenuation and resonance effects were observed at high frequencies depending on the lung parameter set used. The measured attenuation around 4 Hz is particularly critical, as many sleep therapy devices use this frequency for apnea classification.

Endometriosis is a debilitating condition that causes severe pain and infertility, among other morbidities. It affects 10% of reproductive-aged women worldwide and poses a significant public health challenge due to its prevalence and the lack of effective treatments and diagnostic tools. The study of endometriosis is restricted by the difficulty in modelling its multifactorial nature. To address this challenge, we developed a microphysiological system capable of reproducing key features of the menstrual cycle and endometriosis. In this publication, we describe such system and proof-of-concept data using simplified in vitro models that validate its use.

Phrenic nerve stimulation is used as an artificial ventilation method to reduce the adverse effects of mechanical ventilation, such as ventilator-induced diaphragmatic dysfunction. Detailed computational models and electromagnetic simulations are used to determine appropriate stimulation parameters. However, tissue parameters such as the electrical conductivity vary widely in the literature. We aim to identify the tissues that have a considerable influence on the nerve activation threshold. Therefore, an anatomically detailed finite element model of the neck with 13 tissue types, including skin, subcutaneous fat, soft tissue, muscle, bone, cartilage, intervertebral disc, nerve, thyroid, esophagus, trachea, blood, and electrodes, was employed. We calculated the electric potential along the phrenic nerve caused by non-invasive stimulation (2 electrodes with a size of 1 cm × 1 cm). The results were modulated by a monophasic stimulation pulse (pulse width 150 μs) and were applied as extracellular potentials to the MRG nerve model of a myelinated axon (diameter 10 μm). The minimum amount of current required to trigger an action potential was defined as the nerve activation threshold. We set iteratively the electrical conductivity of each tissue to the minimum and the maximum values based on the data from the IT’IS foundation. Considerable changes in nerve activation thresholds were observed for the following tissues: muscle (25mA), skin (22mA), soft tissue (18mA), subcutaneous fat (12mA), and nerve (3mA) tissue. Changes in the nerve activation threshold due to changes in skin conductivity occurred due to the bridging effect, which is an unwanted and avoidable effect during stimulation. In conclusion, fat, muscle, nerve, and soft tissue require the most accurate tissue properties and geometric representation within the model.

Left bundle branch block (LBBB) refers to a delayed or blocked impulse transmission through the left bundle branch of the His-Purkinje system and is often associated with structural heart disease. It also has been hypothesized that LBBB in itself can cause dyssynchronous heart failure. We used an electromechanical heart model with an integrated His-Purkinje system to simulate a healthy and a LBBB case. The LBBB affected only the His-Purkinje system and was not accompanied by other cardiovascular abnormalities. Hence, this case is called pure LBBB. For both simulations, we compared electrophysiological, hemodynamical and mechanical readouts. Our findings suggest that pure LBBB does not immediately lead to a severe change in cardiac output. As a consequence of mechanical dyssynchrony, we noticed a 2.5% to 6.7% absolute strain increase in the left ventricle. We hypothesize that this load imbalance causes remodeling over time, leading to a major reduction in ejection fraction. In line with clinical findings, our research indicates that electrical dyssynchrony immediately affects mechanics more than hemodynamics. These findings can provide insight and inform future research into the underlying mechanisms and efficacy of pacing treatments like cardiac resynchronization therapy.

Preterm neonates often require mechanical ventilation due to immaturity of the respiratory system at birth. Physiological Closed-Loop Control (PCLC) has the potential to improve therapy and relieve clinical staff. We present an integrated model of the patient and mechanical ventilation setup for Model-in-the-Loop simulation of PCLC ventilation in neonates. The model uses settings of pressure-controlled mechanical ventilation as inputs and generates simulated measurements of airway pressure, airway flow, and time capnograms as outputs. We compare the model outputs with measurements from the ventilation of preterm lambs. The results are promising and we expect the model to enable a future X-inthe- Loop validation pipeline for PCLC ventilation of neonates.

In the treatment of hydrocephalus, complications caused by shunt occlusion are common and one of the main reasons for shunt failure. With the goal of optimizing the hydrocephalus valves, an in vitro testing method is designed and implemented to mimic occlusion. The main unit is a flow chamber, simulating the cell growth inside the shunt. A cell suspension is circulated through a chamber using a syringe pump system. Cell flow can be modified and tested against different materials, surface properties and shapes. After flow simulation, the cell attachment and viability can be assessed using fluorescence microscopy. The designed test bench facilitates the development process of hydrocephalus valves, concerning choice of materials, surface finishes and inner (flow) geometry. The syringe pump allows for a steady or changing flow circulation, creating the dynamic conditions for investigation of growth of tissue in the valve. The presented flow chamber can be used to investigate the prevention of occlusion in hydrocephalus valves.

The assesment of the safety and efficacy of medical devices can be time-consuming, expensive, and raise ethical concerns. Additionally, the high costs of innovation and lengthy development times may result in increased prices and unequal access for certain patient groups. In silico clinical trials (ISCTs) offer an innovative alternative by utilizing computational modeling and simulation to generate clinically relevant data. A recent draft by the FDA suggests the use of clinical data to support model validation. One example of this is a case study that investigates paravalvular leakage in the context of transcatheter aortic valve replacement.

Endothelial cells (ECs) are constantly exposed to mechanical forces including shear stress and cyclic strain in vivo. This study explored how ECs respond to radial and 3D cyclic stretching using a collagen-based hybrid membrane system and a pneumatic control (MPSstimulus) to mimic dynamic mechanical conditions. Although deformation of the cell substrate was very low and occurred mainly at the polymer mesh that stabilized the collagen matrix, ECs still showed notable changes such as increased stress fiber formation and cell number. These results highlight the sensitivity of ECs to even minor mechanical cues in 3D environments. Overall, the study underscores the importance of integrating threedimensional mechanical forces in the design of in vitro vascular models.

Dynamic causal modelling is a promising tool to quantify hand movement neural control. It portrays the temporal path of signals across a network of distinct motor control areas in the brain. Optimising the input representation for this modelling framework-including defining the input signal and its entry point into the motor-control network- could enhance its ability to explain interactions within the brain. A more precise characterisation of these inputs may also shed light on visuomotor coordination, which is linked to efficient brain-computer interface usage and bionic hand control. This study applied Bayesian selection to determine the optimal input model. Electroencephalography recordings were acquired from three participants who performed wrist extension and wrist flexion movements. Nine candidate input models were compared using the Bayesian selection's logevidence. The selection was evaluated by the spectral variance between each participant’s measured spectrograms and those predicted by the winning model. In all participants, the same input model was selected: an input signal of three gamma functions entering the left and right premotor cortices. This input model provides insights into visuomotor coordination and can improve the performance of a brain-computer interface controlling a bionic hand.

Covered stents are increasingly used in the treatment of aortoiliac occlusions and aneurysms. However, the presence of a polymeric cover may inhibit endothelialization and alter blood flow, potentially increasing the risk of thrombosis. Accurate geometric models of the stent lumen are crucial for computational fluid dynamics (CFD) simulations for thrombosis risk assessment. This study presents a workflow combining silicone casting, μCT imaging, and CFD for thrombosis risk assessment. Applied to a Gore Viabahn VBX stent, the workflow achieved high accuracy with deviations below 0.1mm. CFD simulations identified low wall shear stress regions downstream of cover-induced constrictions, critical for thrombus formation. This approach offers valuable insights for stent design optimization.

Blood gas analyses (BGAs) are used during mechanical ventilation as a diagnostic tool for partial pressures of gases in the blood, but deliver only spot measurements due to their invasive nature. We evaluate an established neonatal pulmonary model on its suitability to estimate partial pressures of oxygen and carbon dioxide to bridge the gap between BGAs. The evaluation is twofold. First, we conduct a comparison to the model’s original evaluation, using five patients with one simulation each, to underline the importance of reproducibility of implementations. Second, we perform an evaluation based on a collected dataset of 11 patients and 452 simulations, comparing the simulated partial pressures of O 2 and CO 2 to BGA measurements. We report considerable deviations from real measurements and propose adaptations to improve the simulation.

The primary cause of death in extremely prematurely born infants is acute respiratory distress syndrome (ARDS). To manage this condition, both invasive ventilation and less invasive methods like continuous positive airway pressure (CPAP) are employed. Recently, there has been growing interest in delivering aerosolized medication directly to the lungs as a non-invasive therapeutic option. However, the effectiveness of such aerosol delivery in extremely premature infants remains poorly understood due to their unique airway anatomy. To address this gap, a three-dimensional model of the upper airway geometry of an extremely prematurely born neonate was segmented from an MRI scan. This model was used to perform computational fluid dynamics (CFD). The analysis aimed to evaluate airflow behavior and aerosol deposition patterns under CPAP ventilation conditions. The results demonstrate velocity and pressure contours that are consistent with previous CFD studies in adults and older infants, while revealing new quantitative insights specific to the narrow and delicate airways of extremely premature neonates. Notably, the simulations capture the presence of a laryngeal jet even at this early developmental stage. Furthermore, particle tracking analyses highlight specific regions prone to frequent aerosol deposition, which may limit the uniform delivery of inhaled medication to the lower lungs. These findings suggest that CFD modeling can play a crucial role in optimizing aerosolbased therapies for this vulnerable population. Further work is needed to validate the results against in-vitro experiments and to expand the model to include the oral cavity, enabling future investigation of oral inhalation strategies.

Individuals with motor impairments often face challenges in interacting with digital devices. Video-based pose detection has emerged as a promising approach for creating accessible human-machine interfaces (HMIs). However, due to the diverse and highly individualized movement capabilities among this population, generic pose recognition systems are often inadequate. This study investigates the feasibility of a personalized hand-pose recognition system for individuals with motor impairments using minimal training data. We implemented a video-based pose classification model using a webcam and the MediaPipe Hands framework to extract keypoints from hand poses. Each participant, regardless of their level of impairment, selected four individual hand poses according to their capabilities, which were then used to train a neural network. The trained models were evaluated in a realtime test application where participants controlled an animated figure using their hand poses. Results from 14 participants demonstrated that most were able to achieve effective control, even in cases of severe motor limitations. Individuals with restricted finger mobility successfully adapted by utilizing wrist and elbow movements. However, participants with spasticity experienced higher misclassification rates due to difficulties in maintaining stable poses. Overall, our findings highlight the importance of individualized pose recognition systems for assistive technology. Future work should explore additional adaptations, such as head or eye-based controls, to further improve accessibility for users with severe motor impairments.

In this paper, we propose a proof of concept for using 6D IMUs to assess the correctness of sit-to-stand (STS) motion, a critical task in post-stroke rehabilitation. Therefore, healthy individuals were instructed to mimic post-stroke movement patterns, providing a relevant model for post-stroke behavior. We demonstrate that using simple kinematic models for all articulated joints is a feasible approach to calculating joint angles and extracting relevant features. Using rule-based and learning-based classification models, we are able to distinguish between multiple error types in the STS motion, with average classification accuracies of 89.78% and 94.03%, respectively. In contrast to many state-of-the-art methods that only provide a binary classification of correct versus incorrect execution, our approach enables the differentiation between various error types. Additionally, it overcomes the limitations of complex, location-dependent camera systems and susceptibility to magnetic disturbances. This highlights the feasibility of this approach for evaluating STS motion, with potential applications in providing automated feedback for post-stroke rehabilitation and supporting the recovery of activities of daily living.

In addition to effective training and optimization of physiological performance parameters, selecting the appropriate sports equipment is key to achieving ideal athletic performance. For Paralympic cross-country sit skiing athletes, this means that the sledge must be tailored to the individual needs of each athlete. An efficient sitting position and an individual sled design tailored to the athlete’s impairment are crucial for the success of para-athletes. However, there is a lack of studies examining the effects of different sitting positions on sledging performance. In this study, a kinematic analysis of three representative positions was performed: knee high (KH), knee low (KL) and neutral (NT). The aim of the analysis was to identify differences in upper body kinematics among six athletes.The results indicate that the subjects had a greater range of motion in the KL position, which allowed for better force transfer onto the poles and thus resulted in significantly improved poling force and thus better forward drive. In contrast, the KH position exhibited a more restricted range of motion in the upper body, which could limit force transfer and their performance. In summary, the findings suggest that the KL position maximizes the upper body range of motion and overall efficiency in poling. This is significant for the selection of the individual sledge position and for fair classification in the sport

Early diagnosis of neurodevelopmental disorders in infants relies on accurate analysis of spontaneous movements. Achieving this requires fast and precise pose estimation methods tailored to infant-specific anatomy and motion. This study evaluates several pretrained YOLOv11-pose models for pose estimation in depth video recordings of preterm neonates and infants using the open source babyPose data set database. The fastest model (YOLOv11n-pose) has a inference time of 0.007 seconds. Considering a previously proposed data split without subject-wise separation between training and testing data, the most accurate model (YOLOv11m-pose) has a median root mean squared distance (RMSD) of 2.15. The median Dice Similarity Coefficient (DSC) and Recall (R) of the joints are 0.85 and 0.86, while the median DSC and R of the joint connections are 0.90 and 0.91. Considering a subject-wise separation of training and testing data, the results noticeably degrade, e.g. to a median DSC and R of the joints of 0.79 and 0.81, while the median DSC and R of the joint connections are 0.75 and 0.79. The present work demonstrates a fast and, copared to the literature, accurate approach to depth-based pose estimation in preterm neonates and infants paving the way for automated movement analysis as a clinical tool for early detection of developmental impairments. Particularly in semiautomated settings where subject-specific annotations can be provided, the results are convining. Regarding the abilities to generalize, more work is required to improve the results.

Adolescent idiopathic scoliosis (AIS) is a threedimensional deformation of the spine typically occuring in adolescents, mainly females. Untreated AIS can lead to the need for surgical intervention or to severe back problems later in life. The conventional treatment of AIS involves the use of custom-made braces, combined with physiotherapy. The therapeutic efficiency of the braces is determined using static X-rays. Dynamic evaluations are based on the experience of orthopedic technicians. However, evidence-based knowledge is needed to compare treatment objectively. Gait analysis offers the potential of dynamically assessing the effect of braces without radiation. To evaluate the evidence-based effect of braces using gait analysis, it is first necessary to identify differences between gait with and without brace. Therefore, in this pilot study, pedobarographic gait analyses of nine AIS patients were examined and evaluated. We comprised 199 dynamic gait parameters extracted from pedobarographic measurements covering spatiotemporal gait parameters, pressure and force distributions of foot zones. Logistic regression analysis was used to classify gait with and without brace. A significant model (p < 0.01) with two significant coefficients (p < 0.05) was revealed covering time to maximum force in relation to segment contact time for left rear and maximum pressure of left toe. Overall classification performance showed an accuracy of 88.9% to distinguish between with and without brace. Our study demonstrates that there are differences in the gait patterns of AIS patients with and without braces. Further studies, including a more extensive data set, may yield evidence-based findings on the effect of braces on dynamic gait patterns to be used in customized braces for the improved effectiveness in therapy.

Optically pumped magnetometers (OPMs) are magnetic field sensors which can be used for different biomedical applications. Due to their modularity and small size, they can be flexibly arranged over the region of interest. For inverse calculations, it is necessary to know their exact location and orientation. In this paper, we compare three different ways to localize OPMs: 3D structured-light scans, a 3D point digitizer and a coil array. We used each modality ten times on a head phantom to assess their consistency. 3D scans provided the most reliable way to localize OPMs with an average difference of sensor distances of just 0.2mm and an average difference in sensor orientations of 0.5∘ in all directions.We also discuss modality-specific advantages.

Hydrocephalus is typically treated through the implantation of shunt systems, which aim to manage the imbalance in cerebrospinal fluid (CSF) production and absorption. Programmable shunt valves (PSVs), such as the Polaris® SPV200, are commonly used in these treatments as they allow for non-invasive adjustments to regulate CSF drainage. This study provides information on the improvement of a validated test setup using the different opening pressures of a programmable shunt valve to open up the hydrodynamic characterization of shunts as a field of application. The testing was conducted under controlled conditions, measuring both flow rates and pressures at different opening pressures. The experimental results were then compared to the target values. The findings revealed slight deviations between the measured performance and the target results. These deviations highlight the potential influence of environmental factors and testing conditions on valve performance. The results suggest that further investigation, including testing under varying temperature conditions, is necessary to optimize the used test setup.

Mechanical ventilation, as performed daily in intensive care units, can lead to ventilator-induced diaphragmatic dysfunction. A reduction in diaphragmatic force generation can occur within the first 24 h. Electrical stimulation of the phrenic nerve can mitigate this effect, but optimal stimulation parameters are largely unexplored. The aim of this work is to build a multimodal measurement and stimulation setup allowing for the test of a wide range of stimulation parameters. Ultrasound, breathing belt, electromyogram, and electrocardiogram measurements are performed simultaneously. The results demonstrate the feasibility of transcutaneous electrical phrenic nerve stimulation, showing observable responses in diaphragmatic movement and breathing belts. The developed system provides a platform for further investigations into the effects of phrenic nerve stimulation.

In upper limb amputees implantation of prostheses equipped with electrodes to innervate respective musculature might improve sensitivity and life quality. Besides different types of prostheses being available, there is still a need for devices that provide more comfort and functionality to patients. Prior to prosthesis implantation magnetic resonance imaging (MRI) or ultrasound are suitable methods for imaging peripheral nerves but are limited in description of nerve dimension. Using X-ray micro-computed tomography (micro- CT) exact dimensions of nerves can be obtained. However, since ex vivo micro-CT analysis on human nerves is rarely possible, the underlying study analyzed whether porcine nerve tissue is comparable to human nerve tissue and thus suitable for nerve electrode implantation studies. Therefore, micro-CT analysis was performed at 10 different proximal to distal sections of porcine Nervus ulnaris, Nervus medianus and Nervus radialis to determine the diameters of nerves and fascicles (n = 1). The diameters of the porcine Nervus ulnaris, Nervus medianus and Nervus radialis were 2.99 ± 0.28 mm, 4.40 ± 0.38 mm and 7.89 ± 0.98 mm, respectively. The diameters of the human Nervus ulnaris, Nervus medianus and Nervus radialis were measured at 10 different regions using a caliper and were 2.93 ± 0.71 mm, 3.65 ± 1.06 mm and 2.94 ± 1.11 mm, respectively (n = 2). Fascicular diameters of porcine Nervus ulnaris, Nervus medianus and Nervus radialis were 0.22 ± 0.09 mm, 0.26 ± 0.06 mm and 0.29 ± 0.07 mm, respectively. Our preliminary results showed that nerve diameters were similar between pigs and humans. Espcially, further analyses of porcine Nervus ulnaris, Nervus medianus and Nervus radialis are necessary to conclude whether porcine nerve tissue is suitable for electrode implantation studies.

Microelectrode arrays (MEAs), a widely used class of neural implants, are susceptible to in vivo degradation despite using biocompatible materials like polyimide (PI). While PI has been extensively studied for its biostability, the role of oxidative stress, particularly from hydrogen peroxide (H2O2), a reactive oxygen species (ROS) released during foreign body reactions (FBR), remains underexplored. This study investigates the effects of in vivo environments, with a particular focus on H2O2 exposure, on the PI layer of MEAs. Scanning Electron Microscopy (SEM) and Time of Flight-Secondary Ion Mass Spectroscopy (ToF-SIMS) are used to study the structural and molecular changes, respectively. The findings indicate that reactive accelerated aging (RAA) mimicking in vivo conditions do not alter PI at the molecular level. Surface modifications, primarily due to oxidative degradation, were observed; however, the underlying material remained intact, confirming PI's durability for long-term implantable devices.

Magnetoencephalography (MEG) using optically pumped magnetometers (OPMs) offers several advantages over electroencephalography (EEG) and superconducting quantum interference device (SQUID)-based MEG, including improved spatial resolution and versatility. In this study, we investigate the potential of OPM-based MEG for multi-class Brain-Computer Interface (BCI) applications. We classify neuronal activity related to five mental tasks - hand motor imagery (MI), feet MI, mental rotation, mental subtraction, and word association - using data from 12 participants. The goal is to develop a strategy to select the most discriminative task subsets for each user. The data were processed using the filter-bank common spatial patterns (FBCSP) algorithm and linear discriminant analysis (LDA), with classification accuracy evaluated using block-wise cross-validation. The results show that, in all but one participant, classification accuracies exceeded the commonly accepted 70% threshold for reliable BCI control, particularly with three- and sometimes with four-class combinations. Hand MI and mental rotation emerged as the most distinct tasks, followed by word association and feet MI being included in fewer optimal task combinations. The study demonstrates that user-specific task selection is crucial to maximize BCI performance. The presented results are promising, but further development is needed, including the implementation of real-time feedback systems, addressing technical limitations, and refining classification techniques. The findings highlight the potential of OPM-based MEG for future clinical applications, particularly in rehabilitative settings, where its high spatial and temporal resolution could significantly enhance BCI systems.

Understanding how body position changes affect bioimpedance is essential for improving wearable technology. This study assessed the influence of posture and timerelated changes on whole-body bioimpedance in 12 volunteers, who maintained sitting, standing and supine positions for 15 minutes each. Changes in body position influenced bioimpedance. Transitioning from standing to sitting led to a average bioimpedance decrease of -2.97 ± 1.19%, which reversed when returning to a standing position. Moving to a supine position resulted in an initial increase of in average 2.55 ± 1.15%, followed by a further rise to 4.25 ± 0.91% over 15 minutes. These results are consistent with previous studies and confirm that deviations from standard clinical positions influence the results of BIA measurements. The study highlights the need to adapt BIA for non-standard positions to optimize and ensure accuracy. Key factors such as hydrostatic pressure on the change in fluid distribution need to be considered in portable BIA applications and future research should incorporate correction models where appropriate.

As the nursing shortage in Germany continues to intensify, the necessity for user-friendly assistance systems is becoming increasingly important. This paper describes an approach for the automatic and regular recording of vital signs using consumer-grade cameras. The effects of varying environmental conditions and behavioral changes on the accuracy of heart rate and respiration rate measurements are analyzed. The results demonstrate accurate monitoring in controlled settings with varying distances. However, lighting conditions and behavior highly impact accuracy, highlighting the need to tailor algorithms specifically for the respective real-world application.

Fall detection is essential for elderly and disabled individuals, as undetected falls can be life-threatening. Traditional methods use acceleration sensors and neural networks, but body-worn sensors can be intrusive. This study explores neuromorphic cameras for fall detection with minimal data processing and lightweight neural networks. Denoising techniques were applied to event data, followed by statistical analysis to estimate position, velocity, and acceleration. This method produced patterns similar to accelerometers. Neural network architectures were evaluated, from simple one-dimensional convolutional networks to hybrid models combining convolutional layers with Long-Short-Term-Memory units. Training data were generated by converting video-based fall datasets (le2i, MCFD) into event data using the v2etoolbox. Data augmentation resulted in 2,610 samples (1,314 falls, 1,296 daily activities). The best model, a threelayer 1D convolution combined with a two-layer LSTM (hidden size 64, 125k trainable parameters), achieved 97% accuracy. Live inference on streamed videos and a DVXplorer event camera was possible without noticeable delay. Our approach matches state-of-the-art acceleration sensor methods while offering a non-intrusive, real-time monitoring solution, potentially improving response times and user comfort.

With the growing role of telemedicine and artificial intelligence (AI) in healthcare, trust and trustworthiness have become essential considerations - especially for systems relying on medical sensors and patient-specific data. Using Continuous Glucose Monitoring (CGM) as a representative case, this work explores how technical metrics (e.g., accuracy, security, robustness) and subjective user perceptions can be combined to evaluate trustworthiness in the context of Diabetes care, considering different stakeholders’ perspectives. We highlight key risks across the CGM lifecycle and propose a trustworthiness factor that enables dynamic, bi-directional interaction between system assessments and user trust. This approach offers a foundation for making trustworthiness tangible and adaptable in future AI-driven medical applications.

The perfusion diagnostic of the abdomen, especially the liver, is usually done via computer tomography, which implies radiation exposure for the patient. Based on a standard electrical impedance tomograph (EIT) it was investigated, if abdominal perfusion can be visualized. Three subjects attended this study. They did breathe-holding manoeuvres, while wearing the EIT belt in the height of the liver. Respiratory artefacts were removed with a bandpass filter, thus pulsatile signals in the regions of the liver and spleen are visible. Despite its relatively low spatial resolution, abdominal perfusion can be visualised with a standard EIT and opens up new diagnostic opportunities for dynamic perfusion monitoring.

Digital telemonitoring systems offer promising approaches to reducing the burden on nursing staff. In this study, the existing telemonitoring platform COMES® (COgnitive MEdical Systems) was technically further developed for use in outpatient care and tested in a study. Prior to this study, COMES® already received, stored, and displayed vital data from various mobile devices. The system was initially expanded with a Django-based middleware, integrated with the API of the company Withings, and supplemented by a new web app for data visualization designed for nursing staff. The resulting extended system, called DIP (Data Integration Platform), was then tested with various mobile devices (e.g., blood pressure monitor, scale, smartwatch) by 22 care recipients over a period of five weeks. On average, the nursing staff rated the usefulness of the DIP as neutral, primarily due to challenges in integrating it into their daily routines. In contrast, the care recipients rated the system’s user-friendliness as good to very good. The results highlight the potential of telemonitoring systems and the need for better integration of such systems into care processes.

This paper presents the development of a humanmachine interface (HMI) and an external human-machine interface (eHMI) for an autonomous patient transport system (APTS). The idea of the APTS is to increase the efficiency of intrahospital patient transport and to reduce the physical stress on healthcare workers by autonomizing the transport of patients in hospital beds. The eHMI enables interaction with external persons, while the HMI supports direct patient communication. The development incorporated ergonomic and hospital-specific requirements. The eHMI and HMI were evaluated in an online study, assessing their usability and acceptance. The evaluation yielded a positive outcome in terms of perception and acceptance on the APTS.

We present an intuitive, real-time control system for a 6-axis robotic arm, designed to support assistive applications in biomedical contexts. Using four MARG-based inertial sensors placed on the arm and hand, human motion is captured and mapped to robot kinematics via quaternion-based sensor fusion and a two-step calibration process. The resulting end-effector pose is scaled to match the robot’s workspace, while EMG signals enable control of a robotic gripper through thumb-index gripping gestures. The system enables direct, gesture-based control without extensive training, making it accessible for users with limited mobility but preserved upper limb function. Evaluation results confirm stable, low-latency operation during pick-and-place tasks, with users intuitively compensating for minor sensor inaccuracies through visual feedback. This approach demonstrates strong potential for assistive robotics in home or care environments, particularly where conventional control methods are unsuitable.

In robotic ultrasound the probe is attached to the robot’s end effector using custom-developed probe holders. To enable successful and safe scans, high reachability of the target poses as well as precise probe contact wrench control are crucial. This study evaluates how different probe holder geometries influence the reachability as well as the accuracy of the internal wrench estimation model of the KUKA LBR iiwa 14 R820 performing US scans. It could be demonstrated that a probe, which is attached to the end effector with a small offset and 90° rotation increased the mean reachability by 16%. Moreover, mean Euclidian force estimation errors were reduced by 22% and mean Euclidian moment errors by 35%, compared to a typical tool geometry of a straight end effector extension. Thus, the results show a promising approach to improve the accuracy of the robot wrench estimation in medical applications.

Cable-Driven Parallel Robots (CDPR) have gained increasing attention in human-machine-interaction, particularly in haptics and rehabilitation. Despite their different objectives, both fields share fundamental requirements such as safety, configurability and precise force control. This review provides an overview of existing CDPR implementations designed for human interaction, categorizing and analyzing systems based on their design principles and kinematic properties. Key motivations for using CDPR in these applications are discussed, along with current limitations and promising directions. The findings indicate that CDPR offer an underrated yet promising solution for human-machine-interaction.

In this paper, different welding techniques for manufacturing pneumatic artificial muscles (PAMs) for the use in soft everting robots for colonoscopy are presented. The focus is on developing reproducible prototyping methods for manufacturing robust and cost-efficient PAMs. Two welding techniques, including laser welding and contact welding, are analyzed for their effectiveness in creating durable and scalable PAMs. Preliminary results show that laser welding offers high precision but inconsistent airtightness, crucial for PAM functionality. In contrast, contact welding, though slower, provides more reliable airtight seals and better reproducibility. Optimized contact welding settings balance weld seam thickness and production time. Further refinements are needed to improve speed and seam quality, with potential for large-scale production. These findings offer valuable insights for scalable PAM manufacturing in soft robotics.

To ensure the quality of newly developed medical devices, rigorous testing is essential. Conventional manual testing is a time-consuming, costly, and non-standardized process. Therefore, automation and robotics play an important role. This work presents the development of an automated test stand utilizing a robot arm to conduct performance tests on vibroacoustic sensing units, serving as navigation devices for lumbar puncture. Following a series of tests, the results obtained from robotic testing are compared with those from manual testing, which serves as a reference method. Both testing methods show identical outcomes. However, robotic testing demonstrates significantly improved efficiency

Abdominal aortic aneurysms (AAAs) pose a significant health risk, particularly in the elderly population, due to the potential for life-threatening rupture. Conventional 2D ultrasound surveillance and guidelines for surgical intervention based on aneurysm size have limitations, necessitating a more individualized approach to assessing rupture risk. This study investigates the overall accuracy of a novel robotic ultrasound system using a single transducer on a collaborative robotic arm. High quality phantoms were used to test the accuracy of the 3D resolution. The system demonstrated promising lateral reconstruction accuracy for large vessels, with dimensions close to the true vessel size of 6 mm. Some slight deviations existed for the reconstructed longitudinal vessel dimensions due to vessel compression by the probe. Future work will focus on optimizing the reconstruction algorithm and system setup, with plans for testing on abdominal phantoms and human subjects to ensure broader applicability and reliability in clinical settings.

Automatic recognition of surgical phases plays a critical role in enabling intelligent, context-aware support systems during operative procedures. The inherent variability of surgical techniques and intraoperative conditions makes precise surgical phase recognition (SPR) a challenging task. This study explores ensemble learning as a strategy to improve phase recognition performance using a learnable fusion approach. A selection of state-of-the-art deep learning models was trained and tuned to capture complementary aspects of the task. This resulted in 14 base models with varied backbones and parameter settings. To aggregate the output probabilities of the base models, a lightweight fully connected network, referred to as StackingNet, was designed as a metamodel capable of learning to generate final predictions from their outputs. This approach outperformed the best individual base model within the respective ensemble in 14 out of 15 ensemble configurations, achieving a maximum F1-score improvement of 3.3 %. These results demonstrate that learnable ensemble fusion can significantly enhance accuracy of surgical phase recognition, highlighting its potential in the development of intelligent surgical assistance systems.

We present a novel approach to produce medical oxygen in low-resource settings at maximal energy efficiency. The oxygen concentration and the required flow can be configured independently by the user and are adjusted automatically by a multivariable feedback controller which exploits the available degrees of freedom, namely the rotational speed of the compressor motor, the opening of a valve restraining outflow, and the switching time of the pressure swing adsorption process. Extensive experiments prove that substantial energy savings can be achieved over a wide range of operating points. The system is suitable both for supplemental oxygen treatment and continuous positive airway pressure applications.

This study introduces an innovative approach to enhance surgical training using the Lubecker Toolbox simulator. Addressing the limitations of current simulators that rely on unrealistic artificial tissues, we developed custom 3Dprinted fixtures to integrate biological tissue (pig feet) into the training environment. The fixture securely holds the biological specimen, while a specially designed stencil enables trainees to create uniform incisions for consistent practice and evaluation. User feedback indicated a significantly improved training experience due to the enhanced realism and haptic feedback provided by the biological tissue. This research highlights the potential of additive manufacturing to create tailored solutions that address specific needs in medical education, ultimately leading to more effective surgical training.

Efficient detection of surgical instruments in operating room (OR) settings is critical for advancing surgical workflow management systems. While deep learning-based object detection has achieved promising results in minimally invasive surgery, the detection of instruments on the instrument stand during open procedures remains underexplored. This study introduces two annotated datasets simulating realworld OR scenes, featuring instruments from different manufacturers. We evaluate two detection approaches: a standalone YOLOv8x model and a custom two-stage pipeline that combines a YOLOv8x with a ResNet-34 model. Both models were trained and evaluated on the first dataset and subsequently tested on a second dataset containing instruments from different manufacturers to assess cross-manufacturer generalizability. On the first dataset, the two-stage pipeline achieved a mAP50 of 99.2%, outperforming the standalone YOLOv8x model (98.4%) by improving detection accuracy of stacked instruments. However, both models exhibited notable performance drops (YOLOv8x: 75.9%, pipeline: 78.4% mAP50) when applied to instruments from different manufacturers, highlighting the impact of inter-manufacturer variability in instrument morphology. Our findings emphasize the need for more comprehensive, multi-source datasets to enable robust and generalizable instrument detection solutions in diverse surgical environments.

Artificial Intelligence (AI) continues to transform medical imaging and surgical assistance systems, particularly in laparoscopic surgeries. Accurate recognition of surgical instruments across space and time is vital for workflow analysis and real-time decision support. This study presents a spatiotemporal transformer-based approach for multi-label surgical tool recognition in laparoscopic videos. We fine-tune the TimeSFormer network to capture spatial and temporal dependencies across video frames. To address challenges like class imbalance and visual occlusion, we incorporate a targeted data augmentation pipeline, balanced batch sampling, and Focal Loss. A new background masking technique further enhances model focus on tool regions by blurring irrelevant textures. Evaluated on the Cholec80 benchmark, our model achieves a mean Average Precision (mAP) of 96.3%, outperforming prior baselines. Attention heatmaps confirm effective tool tracking, underscoring the promise of spatio-temporal transformers in surgical AI.

Gait analysis provides insights into human motion by examining how individuals walk. However, the high cost associated with gait centers prevents conducting gait analysis regularly, reducing opportunities for early detection and prevention of gait-related issues before pain or injuries occur. The approach presented in this paper integrates low-cost yet computationally powerful hardware with signal processing and machine learning to develop a wearable sensor node placed at the pelvis that continuously collects gait data, providing personalized gait analysis. By positioning the wearable at the pelvis, gait asymmetries can be captured accurately. The approach is validated in a laboratory experiment with 15 participants walking on a treadmill and verified in a free-moving environment. Results indicate that the wearable detects gait asymmetries effectively, enhancing applicability in both clinical and non-clinical settings, supporting rehabilitation and preventive care in a cost-efficient manner.

Sleep apnea is a common chronic sleep-related disorder which is known to be a comorbidity for cerebroand cardio-vascular disease. Diagnosis of sleep apnea usually requires an overnight polysomnography at the sleep laboratory. In this paper, we used a wearable device which measures reflectance photoplethysmography (PPG) at the wrist and upper arm to estimate continuous SpO2 levels during sleep and subsequently derive an oxygen desaturation index (ODI) for each patient. On a cohort of 170 patients undergoing sleep apnea screening, we evaluated whether this ODI value could represent a surrogate marker for the apnea-hypopnea index (AHI) for the diagnosis and severity assessment of sleep apnea. As the ODI was simultaneously obtained at the fingertip, upper arm and wrist, we compared ODI diagnostic performance depending on the measurement location. We then further evaluated the accuracy of ODI as a direct predictor for moderate and severe sleep apnea as defined by established AHI thresholds. We found that ODI values obtained at the upper arm were good predictors for moderate or severe sleep apnea, with 86% accuracy, 96% sensitivity and 70% specificity, whereas ODI values obtained at the wrist were less reliable as a diagnostic tool.

In this work, we present a portable, open-source data logger with fully enclosed hardware that supports modification and operation under demanding conditions such as 100% relative humidity and exposure to chemical disinfectants. Based on a Raspberry Pi Zero W (RPI), the system manages BLE communication with various sensors to extract, analyse and store biomedical data during runtime. Its architecture is extensible with medical-grade, consumer-grade and custom-made hardware for multisensory applications. A web interface enables configuration and data export, making the device suitable for biomedical research. As an example, a Polar H10 sports chest strap is used for ECG- and acceleration (ACC) measurements.

Recent advances in deep learning enable estimation of a patient’s "ECG age" from standard short-term 12-lead electrocardiogram (ECG). This study introduces sequential ECG age predictions in continuous 24-hour Holter ECG recordings of chronic heart failure (CHF) patients. Using publicly available data from the MUSIC study, we analyzed data from 869 CHF patients, assessing both the differences (predicted ECG age vs actual chronological age) and dynamic -including variability and entropy-based complexity- to characterize temporal fluctuations over 24h. After automated removal of segments based on signal-to-noise ratio (SNR), 222 deceased patients were matched 1:1 with 222 surviving patients by sex, NYHA class, and age. While the predicted ECG age is similar across CHF patients, our findings indicate lower variability but increased complexity (approximate and sample entropy) in deceased compared to surviving patients, suggesting more irregular predicted ECG age dynamics among those who experienced adverse outcomes. Our results suggest that longitudinal evaluation of predictions from an end-to-end deep learning model can uncover subtle temporal dynamics potentially valuable for risk stratification.

Photoplethysmography (PPG) is a key sensing modality in wearable devices for monitoring cardiovascular parameters. Despite widespread use, wrist-based PPG measurements often suffer from signal degradation due to anatomical variability and suboptimal sensor placement. This study investigates the influence of sensor positioning on PPG signal quality at the dorsal wrist. A custom-designed sensor array comprising eight photodiodes and multiple green LEDs was employed to perform synchronized multi-position recordings in 20 subjects. Signal quality was quantified using a signal-to-noise ratio (SNR) surrogate derived from bandlimited root mean square values. Results demonstrate that a centrally located distal photodiode (PD2) consistently showed the highest SNR across subjects. Configurations utilizing additional LEDs moderately enhanced signal quality, while combining photodiodes increased signal amplitude without corresponding improvements in SNR. These findings highlight the critical role of sensor placement and emitter configuration in optimizing PPG acquisition.

Multi-energy reconstructions have become an important research field in computed tomography in recent years. Since modern reconstruction and postprocessing techniques often employ deep learning strategies, there is a high need for large, diverse and adaptable multi-energy datasets. Therefore, this work proposes a straightforward pipeline for the generation of multi-energy cone-beam CT projection data based on the established XCAT software phantom with arbitrary desired X-ray spectra. We evaluate the effort and time required for dataset generation and utilize the generated data for model-based iterative reconstruction exemplarily. This approach provides an understanding of the current pipeline’s bottlenecks while demonstrating its suitability in producing high-quality projection datasets and reconstructions. Thus, we contribute to open knowledge on generation of large multi-energetic CT datasets for deep learning purposes.